Students First SF
San Francisco public schools · updated November 16, 2026

More than half of SFUSD's tested students are below the State math standard. Nearly half are below in reading.

In spring 2026, 54.2% of tested students were below the State standard in math and 46.5% in reading. SFUSD is the San Francisco Unified School District, which runs our public schools. The State tests grades 3 to 8 and 11.

Students First SFMath, spring 2026

Math: fewer than 5 in 10 students meet the State standard. For low-income students it is about 3 in 10.

SFUSD students meeting the State math standard, out of 10

State tests, spring 2026, grades 3 to 8 and 11 · district-run schools · group averages, not limits on any child

Show data table

These are group averages, not limits on any child, and they don't show why groups differ. Each filled dot is one student in ten. Low-income is the State's "socioeconomically disadvantaged" group. Two groups with under 300 tested aren't charted. See Method.

Sources

State test results, SF file, 2026; SFUSD on the State's results site, 2025-26; State student group codes.

Students First SFReading, spring 2026

Reading: just over 5 in 10 students meet the State standard. For low-income students it is about 4 in 10.

SFUSD students meeting the State reading standard, out of 10

State tests, spring 2026, grades 3 to 8 and 11 · district-run schools · group averages, not limits on any child

Show data table

These are group averages, not limits on any child, and they don't show why groups differ. Each filled dot is one student in ten. Low-income is the State's "socioeconomically disadvantaged" group. Two groups with under 300 tested aren't charted. See Method.

Sources

State test results, SF file, 2026; SFUSD on the State's results site, 2025-26; State student group codes.

Our position · a value judgment, not a data claim

We read these test results as a red flag. We believe San Francisco should put its children and public schools first. Where the data shows money is not producing results, we would rather it go to students, even if that means spending less on homelessness and drug response.

Students First SFMath, spring 2026

Math: SFUSD is above California overall. Four of six racial and ethnic groups are below the same groups statewide.

Meeting the State math standard: SFUSD and California

SFUSDCalifornia

State tests, spring 2026, grades 3 to 8 and 11 · charter schools left out on both sides · group averages

Show data table

Both sides leave out charter schools. These are group averages, not limits on any child. They don't show why groups differ, and the mix of students inside each group differs between San Francisco and the State. See Method.

Sources

State test results, SF file, 2026; SFUSD on the State's results site, 2025-26; California's non-charter schools on the State's results site, 2025-26 (choose a student group); State student group codes.

Students First SFReading, spring 2026

Reading: SFUSD is above California overall. Four of six racial and ethnic groups are below the same groups statewide.

Meeting the State reading standard: SFUSD and California

SFUSDCalifornia

State tests, spring 2026, grades 3 to 8 and 11 · charter schools left out on both sides · group averages

Show data table

Both sides leave out charter schools. These are group averages, not limits on any child. They don't show why groups differ, and the mix of students inside each group differs between San Francisco and the State. See Method.

Sources

State test results, SF file, 2026; SFUSD on the State's results site, 2025-26; California's non-charter schools on the State's results site, 2025-26 (choose a student group); State student group codes.

The ask: $15.5M a year in new money, plus protection for school money the City already promised.

See the ask Look up your school
Students First SFMath, spring 2026

For scale: Part 3 of the ask would pay for about 4,000 to 8,000 tutoring seats (est.). 12,583 tested students are below the State math standard.

One square is about 126 students below the math standard

  • Seats at $2,500 a student: about 4,000
  • At $1,200 a student: up to about 8,000
  • The rest of the 12,583: about 4,600 to 8,600

A scale comparison, not a count of who is served (est.) · State tests, spring 2026 · district-run schools

Part 3 serves students below the reading or math standard, starting at the highest-poverty schools, and some students are below both. In reading, 10,812 tested students are below the standard, and the same seats would be about 37% to 74% of them (est.). Insights counts a narrower group: 7,846 low-income students below the reading standard, all of whom Part 3 could reach at $1,200 each (est.). Grades without State tests aren't counted. See Method.

Sources

State test results, SF file, 2026; SFUSD on the State's results site, 2025-26; The ask, Part 3; What more money would do (tutoring cost per student).

About 1 cent of each City dollar reaches SFUSD (est.).

The City budget, FY26-27, $16.85B

State formulas and grants set about 70% of SFUSD's revenue in FY26-27. FY26-27 means July 2026 through June 2027.

See the data: Four facts to read with the City budget numbers

Our view: San Francisco should put students first, and there is money it could move.

$167.9MCity money to SFUSD, FY26-27 (est.), about $3,360 per student
$832.7MHomelessness, Department of Homelessness and Supportive Housing (HSH) budget, FY26-27
$698.2MBehavioral health, which includes mental health, DPH division budget, adopted FY26-27

FY26-27 is the budget year from July 2026 through June 2027.

Read these numbers with four facts in mind

State formulas and grants set about 70% of SFUSD's revenue in FY26-27. About $329M of the formula is local property tax, so State cash is about 46% and all local money about 50% (est., SFUSD general fund).

The $16.85B includes the Airport and water and power, which run on their own revenue, and Muni, which also gets about $672M a year in voter-mandated City funding (FY26-27 projection).

On the FY26-27 General Fund budget of $7.60B, the same money is about 2.2 cents per dollar (est.).

Everything else also includes DCYF ($372.8M) and Early Childhood ($342.6M), which fund children's programs outside SFUSD's own budget. Some of the SFUSD money above passes through DCYF.

Budget data as of October 31, 2026.

This is the coalition's view. Every section below is built to be checked on its own terms, including the toughest questions.

Sources

City money to SFUSD is the Public Education Enrichment Fund share ($107.6M), the Student Success Fund ($45.0M), and the Public Education Services Baseline ($15.3M), divided by about 50,000 district-run students. These three amounts did not change in the final budget (Ordinance 140-26). It leaves out the City-levied Fair Wages for Educators parcel tax ($57.3M); with it, about 1.3 cents (est.). HSH's $832.7M is before citywide transfer adjustments. On the same net basis as the $16.85B total it is $803.6M, 4.8%. The Budget Book's $134M is DCYF's whole Public Education fund: $116.3M PEEF, $16.5M PEEF baseline, and $1.45M soda tax. The PEEF line passes $107.6M because it also spends unspent prior-year balances. On those lines the adopted budget appropriates about $177.8M, about 1.1 cents (est.). Controller Revenue Letter, June 2026 (Exhibit 4-1, p. 33: PEEF to SFUSD $107.6M, Student Success Fund $45.0M, Public Education Services Baseline $15.3M; p. 38: Fair Wages parcel tax $57.3M; p. 8: November 2026 measures); HSH Budget FY26-28; Mayor's Proposed Budget Book (PDF p. 12: self-supporting enterprises; p. 125: SFUSD pass-through and prior-year PEEF balances); State Budget Overview for Parents, FY26-27; SFUSD 2026-27 budget, Form 01 (State cash and local shares); Controller March 2026 Update, Appendix E (Muni); Ordinance 140-26, the final budget (PDF p. 13: $16.85B, $7.60B; p. 17: HSH $832.7M, DCYF $372.8M, Early Childhood $342.6M; p. 68: DCYF's Public Education fund; p. 178: behavioral health $698.2M); Controller budget data (HSH $803.6M net of citywide transfer adjustments); SFUSD Facts at a Glance, 2025-26 (50,028 students, not counting charter or county schools); SF Elections, Nov 3, 2026 results (preliminary, November 10 count); Charter Sec. 13.101 (three school board seats in 2026; terms start January 8); Ballot measures, Nov 2026; Board budget hearing calendar, 2026 (the Mayor's budget came June 1; the 2027 date follows this pattern); Mayor's proposed interim budget ordinance, June 2026 (compare its PEEF and Student Success lines with the final budget). All sources and method.

2 of 14 · Insights

Seven things the data shows.

From State, SFUSD, and City records. Where we compare, we don't claim cause. Low-income is the State's "socioeconomically disadvantaged" group. Reading is its English language arts test.

Recovery split by income. Students who aren't low-income are back near 2019 in reading. Low-income students are 5 points below.

Share of SFUSD students meeting the reading standard

There were no tests in 2020. Only 72 SFUSD students took the reading test in 2021, so it isn't shown.

Show data table

Math shows the same split: 1.6 points below 2019 for students who aren't low-income, 7.2 for low-income students.

District rows, which leave out charter schools. The State's cutoff scores haven't changed since 2019. Read 2024 with care: low-income was a larger share of students tested that year. This doesn't show why.

Sources

State test results, SF file, 2019; State test results, SF file, 2022; State test results, SF file, 2023; State test results, SF file, 2024; State test results, SF file, 2025; State test results, SF file, 2026; State test results, SF file, 2021; State, Understanding Results, Smarter Balanced, 2025-26; CDE, 2024 assessment data release FAQs; CALPADS Update Flash 260, Sept 18, 2023.

Low-income is not one group. Reading runs from 16% to 62% among low-income students, 2026.

Low-income students meeting the reading standard, 2026

Show data table

Our view: help should follow each student's need.

These are group averages, not limits on any child, and they don't show why groups differ. Two groups with under 300 tested aren't charted: Native Hawaiian or Pacific Islander (119, 20.2%) and American Indian or Alaska Native (24, 37.5%). The Two or more races group about doubled in the State's files from 2019 to 2022, so its 2019 figure isn't like for like.

Sources

State test results, SF file, 2026; State test results, SF file, 2019; State student group codes (200 to 207 are race groups among low-income students); SFUSD data dashboard FAQ (State definition of socioeconomically disadvantaged).

Most students who need help aren't at the highest-poverty schools. 60% of low-income students below the reading standard attend other schools, though about two-thirds attend schools over 60% low-income (est.).

Low-income students below the reading standard, by school low-income share, 2026 (est.)

Show data table

Part 3 of the ask starts at schools over 75% low-income and follows students below the standard elsewhere. Tutoring all 7,846 would cost about $9.4M to $19.6M a year. Part 3's tutoring money, about $9.9M a year, could tutor about 4,000 of them at $2,500 each, or all of them at $1,200 each (est.).

Among low-income students, 58% meet the standard at schools under 40% low-income, and 23% at schools over 75%. Student mix accounts for about a third of that gap (est.). District-run schools, tested grades only (3 to 8 and 11).

Sources

State test results, SF file, 2026; State census day, 2025-26 (low-income share); EdSurge, Oct 2026 (press; it credits Annenberg, and we did not find the range in the paper it links).

At the highest-poverty schools, fewer students meet the math standard, whether or not the State counts them as low-income. Among students not counted as low-income, 31% meet the math standard at schools over 75% low-income, against 75% at schools under 40%. For low-income students, it's 23% and 47% (2026).

Students meeting the math standard at elementary and K-8 schools, 2026, by the school's low-income share

Show data table

Our view: at schools over 75% low-income, about three in four tested students are below the math standard, so Part 3 starts there and serves students by score, not income. More than half of the students below the math standard at these 68 schools attend other schools (est.), so Part 3 follows them.

Each bar pools students across schools: those meeting the standard divided by those tested. 68 district-run elementary and K-8 schools with at least 100 students; tested grades 3 to 5, or 3 to 8 at K-8 schools. K-5 schools alone show the same pattern. Only 604 students not counted as low-income were tested at schools over 75% low-income, about 26 a school. They may sit closer to the income line than the same group elsewhere, and results vary widely from school to school. These are group averages, not limits on any child, and they don't show why. Who attends each school differs, including through school choice.

Sources

State test results, SF file, 2026; State census day, 2025-26 (enrollment and low-income share). Method: Sources and method.

Chronic absence is highest where scores are lowest. At the typical school over 75% low-income, 44% of students are chronically absent, against 16% at the typical school under 40%.

Median chronic absence by school low-income share, 2025-26

Show data table

One percent more attendance is worth about $6.7M a year to SFUSD at full effect, though the State money follows slowly (est.).

Absence goes with low scores, but student mix explains more, and absence adds little once mix is counted (est.). This doesn't show that absence causes low scores. Chronically absent means missing at least 10% of school days. 2025-26 includes the four-day February 2026 strike.

Sources

State test results, 2026; State chronic absence, 2025-26; State census day, 2025-26; SFUSD 2026-27 Recommended Budget, PDF p. 5 ($14,812 per unit of attendance); State Current Expense of Education, FY24-25 file (attendance of 45,086, San Francisco Unified row); Education Code 42238.05 (three-year attendance average; public mirror); State chronic absence, 2024-25; The Frisc, Feb 2026 (press).

Special education costs SFUSD more than the City gives it. It costs $363.0M (FY25-26). SFUSD's unrestricted money covers $184.4M of it (FY26-27 budget). City money is $167.9M, or $225.2M with the Fair Wages parcel tax (est.).

Dollars a year, on one scale

Show data table

For scale: the State's latest special education rate increase is worth roughly $15M a year to SFUSD, about the size of the ask, and the 2028 parcel tax vote protects about $53M a year (est.).

Years differ, and rows overlap: most bus cost is coded to special education. Many of these services are required by student plans. QTEA ends June 30, 2028 unless voters renew it, and the FY28-29 gap assumes they do.

Sources

SFUSD 2026-27 Recommended Budget: PDF p. 4 (State special education rate, roughly $15M), PDF p. 7 ($184.4M), PDF p. 8 (assumes QTEA is renewed in 2028), PDF p. 22 (parcel taxes, $110.1M), Form MYP, PDF p. 131 ($26.2M), PDF p. 160 (QTEA expires in 2028); SFUSD FY25-26 Unaudited Actuals, Form PCR, PDF p. 156 ($363.0M program cost); SFUSD warrant register, June 2026 (one of 12 Board items, all listed under Who gets paid at SFUSD; Zum, $46.6M); Controller Revenue Letter, June 2026: Exhibit 4-1, p. 33 ($167.9M), and p. 38 (Fair Wages, $57.3M); SFUSD, June 2025 ($113.8M in reductions); SPUR on 2008 Prop A (secondary; $198, levied from July 1, 2008, for 20 years).

Not all the money set aside gets spent. In FY24-25 the City allocated $34.8M from the Student Success Fund to SFUSD, including $8.5M added during the year. SFUSD spent $21.7M.

Student Success Fund, FY24-25

Show data table

Spending what's already set aside costs no new dollars. The Fund's annual report cites hiring freezes and contract-related challenges.

Spending fell about $4.9M short of the original $26.5M grant (est.). Uncommitted money goes to a reserve, last reported at $18.1M. We didn't check the report's table against SFUSD's books.

Sources

City contract data, contract 1000032928 ($26,519,360 agreed, $22,084,862.10 paid); Res. 483-24 (first grant, not to exceed $26,519,360); Controller, June 2024, PDF p. 31 ($35.0M in FY24-25); Controller Revenue Letter, June 2026, pp. 27 and 31 (reserve, $18.1M); Student Success Fund 2024-25 Annual Report (DCYF), PDF p. 32 ($34,797,177 allocated, $21,665,719 spent) and p. 34 (reasons); City voucher data, purchase order 000088211602 ($2,211,622.31 pending); Charter text, Sec. 16.131(d)(6), PDF pp. 22-23.

Also worth knowing

Support staff tilt toward high-poverty schools: 82 students per staff member against 199. Teachers tilt less: 17.5 students per teacher against 20.8 (2025-26, est.). This doesn't show whether those staff raise results.

Empty seats cluster there too: 3,210 of 5,587 empty seats, and 12 of the 14 schools under 70% of capacity (est.). SFUSD's new elementary school assignment system gets a final vote April 13, 2027.

Show data table
Sources

Ed-Data teacher FTE (chart 180; this link shows Lowell, and each school has its own chart); Ed-Data pupil services FTE (chart 187; this link shows one school, and each has its own chart); State census day, 2025-26; SFUSD 2023 Facilities Master Plan, PDF pp. 76-81 (2022-23 program capacity); SFUSD school reorganization page (final vote April 13, 2027).

3 of 14 · SF budget in 5 minutes

State formulas and grants set about 70% of SFUSD's revenue, though State cash is about 46% (est.). Voters set aside about $3.09B of the City's budget.

Who pays for SFUSD, FY26-27

City Hall does not run schools, and only part of its budget can move in any one year.

  • SFUSD is a separate government. The Mayor and Supervisors do not run schools, and State fiscal advisers can veto district fiscal decisions.
  • The City's budget is $16.85B. The General Fund is $7.60B.
  • The Mayor proposes the City budget June 1. The Board of Supervisors adopts it about July 31.
Plain-English glossary
SFUSD
San Francisco Unified School District, the separate government that runs public schools.
PEEF
Public Education Enrichment Fund. City money voters set aside in 2004. Two-thirds goes to SFUSD and one-third to preschool.
Student Success Fund (SSF)
City money voters set aside in 2022 for academics and wellness at school sites. DCYF runs it.
Prop C, or Our City, Our Home
A 2018 tax on business gross receipts. Its money can only go to homelessness purposes.
LCFF
Local Control Funding Formula, the State's main formula for school money.
QTEA
Quality Teacher and Education Act, a 2008 parcel tax for teacher pay. SFUSD levies it, and it runs through June 30, 2028. A parcel tax is a flat tax on each piece of property.
HSH
Department of Homelessness and Supportive Housing, the City's homelessness department.
DPH
Department of Public Health. Its behavioral health division covers mental health and substance use treatment.
DCYF
Department of Children, Youth and Their Families, the City's children's department.
BHSA
Behavioral Health Services Act, the State law from Prop 1 (2024). It sets how counties spend State behavioral health money from July 1, 2026.
See the data: Who runs, pays, and decides

Who runs schools

City and CountyMayor and 11 Supervisors
$167.9MCity money, FY26-27 (est.)
SFUSDIts own elected Board of Education

SFUSD is a separate government. The Mayor and Supervisors do not run schools. They can send money through set-asides and grants. State fiscal advisers can also veto district fiscal decisions.

Who pays for schools

State formula and other State money about 70%, other local money 26%, federal about 4%, FY26-27. About $329M of the formula is local property tax, so State cash is about 46% and all local money about 50% (est., SFUSD general fund). LCFF is the Local Control Funding Formula, the State's main school formula.

The City's budget, FY26-27

$16.85B

The General Fund is $7.60B. The rest is largely self-funded enterprises, such as the Airport, and restricted money. Voter baselines, set-asides, and dedicated taxes take about $3.09B (FY26-27 proposed budget), or about $2.08B without dedicated taxes.

Who decides

  1. June 1The Mayor proposes the City budget.
  2. June and JulyThe Board of Supervisors reviews it. Add-backs, set in committee in late June, are small, $41M in 2025 and about $28M in 2026, citywide.
  3. About July 31The Board of Supervisors adopts it.
  4. SFUSD, in JuneSFUSD's Board of Education adopts its own budget, on June 23 in 2026.
Sources

State Budget Overview for Parents, FY26-27 (revenue of $1.38B, SFUSD plus County Office); Controller budget data and Revenue Letter, June 2026, Exhibit 4-1 ($3,085.8M: $1,235.7M General Fund mandates, $844.3M expenditure requirements, $1,005.8M dedicated taxes; and the $167.9M: PEEF to SFUSD $107.6M, Student Success Fund $45.0M, Public Education Services Baseline $15.3M; p. 38: SFUSD levies QTEA) ($16.85B and $7.60B); SFUSD 2026-27 budget, Form 01 (about $329M of the formula is net property tax); Mission Local, Oct 2026 (press), and The Frisc, Feb 2026 (press) (press reports on the State advisers); Mission Local, June 2026 (press) (press report: $41M in 2025) and Sup. Chan's July 2026 column (the budget chair's own account: about $28M in 2026); SFUSD, June 2026 (budget adoption); Charter Sec. 2.100 (eleven Supervisors); Charter Sec. 8.100 (SFUSD is run by an elected Board of Education); Charter text on PEEF and the Student Success Fund (set-asides and grants to SFUSD; PDF p. 8: preschool share; pp. 21-23: Student Success Fund); SFUSD Fiscal Stabilization Plan, June 2024, PDF p. 8, quoting the State's letter (advisers have stay or rescind authority); Mayor's Proposed Budget Book, PDF p. 12 (55% of the budget is self-supporting, mainly the Port, MTA, Airport, and PUC); Board budget hearing calendar, 2026 (June 1 proposal; June 25 committee deliberation; July 28 final vote; July 31 deadline); Charter Sec. 9.100(b) (Board adopts the budget from July 15 to August 1); Controller on Our City, Our Home; SPUR on 2008 Prop A (secondary; $198, levied from July 1, 2008, for 20 years); DPH deck on the BHSA plan, Health Commission, Apr 2026, slide 5. 2027 budget dates follow the prior cycle's pattern.

4 of 14 · The money

Homelessness and substance use get about $954M to $966M in FY26-27 (est.), and most of it pays for housing and shelter.

$954M to $966MHomelessness and substance use, FY26-27, all funds (est.)
$742M to $799MThe City money inside that total, FY26-27 (est.)
About $121MSpecialty addiction treatment, $120.9M in FY24-25

HSH puts about 80% of its two-year budget into housing and shelter. That is about $664M of this total in FY26-27, or about 69% (est.).

Homelessness funding grew 4.6 times after inflation, FY06-07 to FY24-25 (est.). Substance use treatment grew about 1.2 times, about 20%, over the same years, on definitions that differ (est.)

Show data table

Bars and dots show each year's dollars before inflation. FY25-26 is preliminary, and FY26-27 is the adopted budget. FY17-18 to FY19-20 are missing.

See the data: Budget pies and per-person costs

HSH puts about 80% of its two-year budget into housing and shelter, about $664M of this total in FY26-27, or about 69% (est.). Specialty addiction treatment is about $121M.

The bars and dots show each year's dollars before inflation. Combined, the two lines went from about $145M in FY06-07 to about $761M in FY24-25, about 3.2 times after inflation (est.). The City's 2017 accounting change removed this line from the ledger. DPH reports give specialty budgets for later years: $71.4M (FY20-21), $75.6M (FY21-22), $91.3M (FY22-23), $109.1M (FY23-24). FY17-18 to FY19-20 are missing. FY25-26 is preliminary and FY26-27 is the adopted budget.

The City budget by service area, FY26-27, $16.85B

Public Works, Transportation and Commerce includes the Airport, water and power, and Muni, and is 35% of the budget by Controller organization group.

Substance use treatment, FY24-25, $120.9M. Only 3% is youth prevention.

Who gets paid. The top three nonprofits got about 35% of the homelessness department's nonprofit payments, FY24-25.

The top three are Tenderloin Housing Clinic, Episcopal Community Services, and Five Keys.

For the same view of SFUSD, see Who gets paid at SFUSD.

Per person, on a stated basis

City money is about $3,360 per student (FY26-27) and about $3,000 per person in substance use treatment (FY24-25, est.). City money to the homelessness department (HSH) is about $11,700 to $18,900 per person served (FY24-25 actual) (est.), against about $3,360 per student on Charter set-asides (FY26-27 budget) or about $7,230 counting local school money outside the State formula (about $13,800 with local property tax) (est.). HSH pays rent for people it houses, while families pay for their own housing.

See the data: City budget, 20 years

San Francisco's actual spending rose from $6.45B in FY06-07 to $18.18B in FY24-25, 1.7 times after inflation, and real spending per resident rose 60% (est.). The share reaching SFUSD rose from under 0.6% through FY16-17 to about 1% in FY26-27, while homelessness and behavioral health are now about 9%.

Real City spending rose 1.7 times from FY06-07 to FY24-25; nominal spending rose 2.8 times (est.)

Actual spending, all funds, net of the Controller's transfer adjustments. Real = FY25-26 San Francisco dollars (est.). FY25-26 ($19.09B) is preliminary and not shown. Sources: Controller Spending and Revenue (DataSF bpnb-jwfb); BLS CPI-U, San Francisco area.

Real spending per resident rose 60%, from $14,331 to $22,868 (est.)

Real actual spending divided by Census July 1 population for the fiscal year's first year (est.). Population grew 7% over the period; the FY21-22 jump is partly a 7% pandemic population drop. Sources: DataSF bpnb-jwfb; Census county estimates, Vintage 2025; Vintage 2020; 2000-2010 intercensal.

Human Welfare grew fastest, 2.2 times in real terms; Public Protection grew slowest, 1.3 times (est.)

Growth in actual spending, FY06-07 to FY24-25, by Controller organization group

Human Welfare (incl. HSH, HSA, DCYF)Community Health (DPH)Other groups
Show data table

Organization groups are net of transfers, so they add to the City total. Public Works, Transportation & Commerce is 44% of actual spending but 35% of the budget. We read the gap as Airport and PUC capital programs spending multi-year appropriations: in FY24-25, about $2.5B of this group's actual spending sits in a fund category ("Unspecified") with no budget (est.). Sources: DataSF bpnb-jwfb; DataSF Budget xdgd-c79v.

DCYF grew 2.7 times in real terms; Public Works and Juvenile Probation shrank (est.)

Select a column heading to sort. Real growth compares FY24-25 with FY06-07 in FY25-26 dollars; HSH starts in FY16-17.

Department gross (the Appropriation Ordinance convention), which double counts internal service departments and General City Responsibilities; do not add rows. Economic & Workforce Development grew from a $5.5M base. Sources: DataSF bpnb-jwfb; DataSF xdgd-c79v; Mayor's Proposed FY27 and FY28 Budget Book.

Homelessness spending grew 4.6 times in real terms, the fastest major line (est.)

Homelessness core spending rose from $144M to $662M (real, $M)

Actual. Predecessor programs (HSA Homeless Services, DPH Community Support-Housing, Mayor Homeless Services) to FY15-16, then HSH. County cash aid not included. HSH also took programs from MOHCD and DCYF, so the FY06-07 base is narrower and growth is somewhat overstated.

Police overtime ran $75M over budget in FY24-25 ($M)

Object "Overtime." Actual exceeded budget by $48M to $76M every year since FY21-22. FY25-26 actual ($142.2M) is preliminary. Budget data start in FY09-10.

DCYF grew 2.7 times in real terms ($M, est.)

Actual, department gross. DCYF's Children's Fund spending rose from $44.5M (FY06-07) to $247.4M (FY24-25).

Sources: DataSF bpnb-jwfb (department codes HOM, POL, CHF; program codes for predecessor programs); DataSF xdgd-c79v; HSH Budget FY26-28.

In FY26-27, 1.0% of the City budget reaches SFUSD (est.); homelessness and behavioral health take 9.1%

Show all years

SFUSD channel = PEEF two-thirds + Student Success Fund + Public Education Services Baseline (FY06-07 to FY16-17: DCYF's recorded PEEF payments, which lag and understate Charter amounts; the FY06-07 Charter contribution was $20M, with $0.1M recorded). With the City-levied Fair Wages for Educators parcel tax ($57.3M), City-controlled money to SFUSD in FY26-27 is about $225M, 1.3% (est.). About half of HSH's money is Prop C, restricted to homelessness. About four-fifths of behavioral health money comes from outside the General Fund, much of it with legal limits (FY25-26: General Fund $119M of $656M, DPH). So shares describe priorities, not a pool to redirect. Sources: DataSF bpnb-jwfb; Controller Revenue Letter, June 2024; Controller March 2026 Update; Controller Revenue Letter, June 2026; Budget Book.

What it could buy: HSH's FY26-27 budget equals about 5,172 teachers (illustrative)

Illustrative

This shows scale only. It is not a claim the money can legally move; see Who controls the money.

Unit costs are FY26-27 estimates: average teacher salary from the UESF schedule plus SFUSD's 48.8% benefit ratio; social worker at step 5 on the 2026-27 schedule; tutoring at the $1,200 to $2,500 per student that EdSurge reports, crediting Annenberg Institute research. SFUSD had about 2,860 full-time-equivalent teachers on Ed-Data's count, which is not comparable with SFUSD's own count of 2,947, and 99 school-site social workers in FY25-26. Sources: DataSF xdgd-c79v; Ed-Data teacher FTE; SF Standard, June 2026 (press; 99 school-site social workers); EdSurge, Oct 2026 (press; $1,200 to $2,500 per student; it credits Annenberg, and we did not find the range in the paper it links).

The series has four breaks to keep in mind

  • Budgets start in FY09-10. The Budget dataset begins then. The Spending dataset goes back to FY98-99; this site's actual series starts in FY06-07.
  • The July 2017 accounting change removed program detail, so DPH behavioral health, HSA homeless programs and PEEF cannot be isolated after FY16-17, and several department codes changed.
  • FY17-18 Public Works and PUC actuals are blanked. Conversion entries moved about $4B between them; they net out in City totals.
  • FY25-26 actuals are preliminary (books not closed; data as of Oct 25, 2026).
  • Actual spending runs 5% to 25% above the adopted budget each year. We read the gap as mostly multi-year capital and bond spending (est.). Behavioral health is defined by program before FY2017 and by DPH division now, so shares are not like-for-like across eras.

Sources: DataSF Budget metadata; DataSF Spending and Revenue metadata; Controller Spending and Revenue data (rows begin FY98-99, though the metadata says fiscal 2000; program field turns generic from FY17-18; FY17-18 Public Works and PUC conversion entries).

Sources

Controller Annual Report, FY17-18, PDF p. 9 (new financial system launched July 2017); Civil Grand Jury, June 2026, PDF p. 9 and p. 56 (HSH formed from programs at DPH, HSA, MOHCD, and DCYF); Ordinance 140-26, the final budget (PDF p. 17: HSH $832.7M; p. 178: behavioral health $698.2M, and $656.1M in FY25-26); Controller Budget data (DataSF xdgd-c79v: City totals and HSH budgets); Charter text on PEEF, PDF p. 4 (FY06-07 contribution of $20M); DPH deck on the BHSA plan, Health Commission, Apr 2026, slide 11 (behavioral health revenue by source, FY25-26); UESF certificated salary schedules (SFUSD page); SFUSD 2026-27 budget, Form MYP, PDF p. 134 (benefits $359.0M ÷ salaries $735.2M = 48.8%); SFUSD Facts at a Glance, 2025-26 (2,947 TK-12 teachers); Board Resolution 471-25 (FY25-26 Student Success grant, $29.0M); Controller Revenue Letter, June 2026, p. 31 (Student Success Fund reserve, $18.1M).

See the data: City homelessness spending

HSH's budget grew from $224M in its first year to $832.7M in FY26-27. The total homeless count went from 6,858 (2017) to 7,973 (2026). Unsheltered homelessness hit a 15-year low in 2026, and family homelessness rose 15%.

Over half the homelessness budget is housing, and half the money is Prop C

HSH by program, FY25-26, $785.6M

HSH by funding source, FY26-27 and FY27-28, $1.61B

Housing funds 14,498 units and shelter funds 4,438 beds. Sources: HSH budget overview, Jan 2026; HSH FY26-28.

HSH spent 75% to 87% of its budget in FY23-24 to FY25-26 ($M)

On the Controller's net basis, which this site uses for headlines, HSH spent 87% in FY23-24, 75% in FY24-25 ($629.8M of $834.6M), and 80% in FY25-26 (preliminary). The bars and dots here are department gross, before the Controller's transfer adjustments, so FY24-25 reads $640.5M of $846.3M, or 76%. FY26-27 and FY27-28 are adopted budgets. The FY20-21 spike is mostly the first large Prop C appropriation ($295.2M, with $10.9M spent that year) plus a $172M rise in continuing General Fund money during COVID. Sources: Controller budget data; Controller spending data; HSH FY26-28.

Unsheltered homelessness fell to 3,400 in 2026; the total was 7,973

The 2026 count used a changed method. Sources: BLA 2020 audit (2015 to 2019); HSH 2026 count report.

Overdose deaths peaked at 810 in 2023 and fell to 625 in 2025

Calendar years. Sources: OCME 2025 report; OCME 2024 report; OCME 2023 report; OCME 2022 report; OCME 2021 report; OCME 2020 report; DPH, Substance Use Mortality Trends through 2024, Table 2, PDF p. 41 (2017 to 2019 are DPH counts from death records; 2020 on are Medical Examiner counts).

Sources

2024 PIT Count Report, PDF p. 17 (2022 and 2024); HSH deck to the Local Homeless Coordinating Board, June 2026, slide 8 (changes to the 2026 street count).

See the data: Who gets paid

Ten nonprofits were paid 61.5% of the $495.0M HSH paid to nonprofits in FY24-25, and three were paid about 35%. DPH paid about $392M (est.) to behavioral health vendors that year, led by HealthRIGHT 360 at $77.9M.

For the same view of SFUSD, see Who gets paid at SFUSD.

About $0.93B to $0.96B a year reaches outside providers, most of it nonprofits (est.)

Completed vendor payments, not budgets. The dataset leaves out relocation and rehousing and some human services payments, so HSH payments understate HSH outlays. The Civil Grand Jury, using the Controller's Supplier Payments data, counts $487.3M to 73 nonprofits in FY24-25; the gap likely reflects how nonprofits are defined. Sources: Controller Vendor Payments (Vouchers); Civil Grand Jury, June 2026.

Three housing and shelter operators lead HSH payments

HSH, top 15 recipients

NonprofitNot flagged as nonprofit

DPH behavioral health, top 10 recipients (est.)

NonprofitNot flagged as nonprofit (for-profit or public)

Hover a bar for contract titles. "What they do" comes from City contract titles and provider websites: Tenderloin Housing Clinic, Episcopal Community Services, Five Keys, Compass, St. Vincent de Paul, Brilliant Corners, Larkin Street, Conard House, Progress Foundation, Seneca, HomeRise. Year-to-year swings partly reflect a change at DISH, which left its fiscal sponsor, Tides Center, and became DISH SF in 2025. Source: Controller Vendor Payments (Vouchers), data as of Oct 2026.

How the DPH figures are estimated

No open dataset tags DPH payments as behavioral health. We counted all DPH payments to 26 vendors whose DPH contracts are behavioral health by title, plus behavioral-health-titled contracts at other vendors. We left out the UC affiliation agreement, nursing facility contracts, and IT vendors. Vendors with mixed missions and blank titles are undercounted, and DPH's own civil service clinicians are not included.

Homelessness and behavioral health, in total

About $1.29B (HSH actual spending in FY24-25 plus the DPH behavioral health budget for FY25-26, mostly mental health; est., mixed years) and about $1.53B budgeted for FY26-27 (HSH plus DPH's behavioral health division), or $1.63B counting all behavioral health at ZSFG and Laguna Honda (est.). Homelessness plus addiction treatment alone was $0.75B (actual) to $0.97B (budget) in FY24-25 and is about $954M to $966M in FY26-27 budgets (est.); specialty addiction treatment is $120.9M. About $0.93B to $0.96B a year is paid to outside providers (est.), $0.80B to $0.86B of it to nonprofits.

All top recipients, both departments and years

Top 40 vendors per department per year. DPH rows are behavioral-health-attributed estimates. Source: Controller Vendor Payments (Vouchers).

At least five top-40 HSH payees, paid about $51M in FY24-25, are on the Controller's FY25 corrective action list

In the Controller's FY25 monitoring, Tenderloin Housing Clinic and Episcopal Community Services had no findings, and Five Keys received a waiver, which departments grant for strong recent performance. None of the three is on the FY25 corrective action list. Sources: Controller Nonprofit Monitoring FY25; Civil Grand Jury, June 2026; Controller HomeRise audit, April 2024.

What the Civil Grand Jury found, June 2026

    Sources: Civil Grand Jury, "At Scale, At Risk"; Mayor's release, Aug 31, 2026 (HSH re-bid of its full portfolio through May 2029); Axios, Aug 31, 2026 (press).

    Sources

    Board Resolution 272-25 (Tides Center was DISH's fiscal sponsor; contract moves to DISH SF); Community Forward SF, About (formerly Community Awareness and Treatment Services); DPH, Baker Places financial status update, March 2024 (Positive Resource Center was its parent); Crestwood Behavioral Health, About; DPH deck on the BHSA plan, Health Commission, Apr 2026, slide 12 ($795.6M, including ZSFG and Laguna Honda); Ordinance 140-26, the final budget (PDF p. 17: HSH $832.7M; p. 178: behavioral health $698.2M).

    See the data: Dollars per person

    On City money, San Francisco spends about $3,360 per student and about $3,000 per person in substance use treatment (est.). On all funds, a student gets more: about $27,000, against $7,100 to $12,600 per person treated (est.). Homelessness, per HSH client, is shown separately as a third row.

    On City money, a student gets $3,360 and a person in substance use treatment $3,000 (est.)

    Show data table

    City money is about $3,360 per student and about $3,000 per person in substance use treatment; homelessness is a separate line at about $11,700 to $18,900 (est.)

    What it shows

    City money is about $3,000 per person in substance use treatment (FY24-25 General Fund and Prop C) and about $3,360 per student (FY26-27 Charter set-asides). On all funds, students get about $27,000, most of it set by State formulas; substance use treatment is about $7,100 per person treated, or up to about $12,600 counting partly related crisis response, jail health, and prosecution costs (est.). Homelessness is separate: City money to HSH is about $11,700 to $18,900 per person served (FY24-25 actual) (est.), against about $3,360 per student on Charter set-asides (FY26-27 budget) or about $7,230 counting local school money outside the State formula, including parcel taxes (about $13,800 with local property tax) (est.).

    What it does not show

    Years differ: student figures are FY26-27 budgets; substance use and HSH figures are FY24-25. The 17,029 people treated include some seen only in primary care, jail, or hospital, so $7,100 understates cost per specialty client. HSH serves a targeted, high-need group and pays rent; schools serve every child while families pay for housing. The low end divides by all 43,148 people: the 26,634 HSH served while unhoused plus the 16,514 in its housing programs, assuming no overlap. The high end divides by the 26,634 only. The report doesn't say how far the two groups overlap. HSH puts 56% of its two-year budget into housing ($909.0M), most of it, on our reading, keeping current tenants housed.

    Basis and same-year alternatives: see Method. All-local revenue: CDE Budget Overview for Parents, FY26-27.

    Sources

    CDE Current Expense of Education, FY24-25 ($26,762 per unit of average daily attendance); SFFD Budget Book FY25 and FY26, PDF pp. 100, 105, 108, 113 (overdose, crisis, and EMS-6 lines); DPH Jail Health deck, 2025, PDF p. 8 ($55.4M, FY25-26); Budget and Legislative Analyst memo, April 2019, PDF p. 2 (drug charges 20% to 25% of DA caseloads; an old share); HSH Budget FY26-28 (housing $909.0M, 56% of the two-year budget).

    Sources

    $954M to $966M is the HSH FY26-27 budget ($832.7M) plus specialty treatment at about $121M (DPH, FY24-25) to about $133M (the SF Funding Explorer's FY26-27 total of $965.5M less HSH's $832.7M; a secondary site's extrapolation). DPH has not published a FY26-27 treatment figure. $742M to $799M applies HSH's 83% to 90% local share and adds the $50.8M General Fund and Prop C share of treatment (mixed years). After-inflation figures are FY25-26 San Francisco dollars. HSH Budget FY26-28; DPH Treatment on Demand FY24-25 (PDF pp. 8-9: $120.9M, General Fund $39.9M, Prop C $11.0M, service types, $109.1M in FY23-24; p. 17: 17,029 people); DPH Treatment on Demand FY22-23, PDF p. 14 ($91.3M); DPH Treatment on Demand FY21-22, PDF p. 12 ($71.4M and $75.6M); Controller spending data; Controller budget data; Controller vendor payments; Bureau of Labor Statistics, SF-area consumer prices; Controller Revenue Letter, June 2026 (Exhibit 4-1, p. 33: the $167.9M); State Budget Overview for Parents (local revenue outside the formula, $361.6M); SF Funding Explorer (secondary site; HSH budgets plus DPH specialty treatment, a best estimate); HSH budget and funding overview, May 2023, PDF p. 3 (FY22-23: 83% local); Controller 2025 Homelessness Needs Assessment, PDF p. 7 (26,634 unhoused people served in 2024) and p. 20 (16,514 people in HSH housing programs); SFUSD Facts at a Glance, 2025-26 (50,028 students, not counting charter or county schools); SFUSD 2026-27 budget, Form 01, PDF pp. 19-20 (about $329M of local property tax in the formula).

    5 of 14 · What voters already decided

    Since 2004, voters have earmarked money for both schools and homelessness, which limits what the Mayor and Board can move.

    $3.09BCitywide voter baselines, set-asides, and dedicated taxes, FY26-27 proposed budget. About $2.08B without dedicated taxes.
    50.5%Share of the homelessness department's FY26-27 and FY27-28 budget from Our City, Our Home, $813.9M of $1.61B
    2/3Board of Supervisors vote needed to change Our City, Our Home, and only to further its purposes

    Each earmark protects a priority voters chose, and each one leaves the Mayor and Board fewer dollars to move.

    1. 2004PEEF (Prop H): $161.4M for public education in FY26-27, through June 30, 2041.
    2. 2008QTEA (Prop A): teacher parcel tax, ends June 30, 2028 unless renewed.
    3. June 2018Educator parcel tax (Prop G), replaced by Fair Wages (Prop J): $57.3M in FY26-27.
    4. Nov 2018Our City, Our Home (Prop C): business tax, homelessness uses only.
    5. Nov 2022Student Success Fund (Prop G): $45.0M in FY26-27. The City may freeze or cut it, never below $35M.
    6. March 2024Prop F: treatment requirement for county cash aid. Sets aside no money.
    7. 2024State Prop 1: new rules for county behavioral health money, fewer dollars for prevention.
    8. June 2026"Overpaid CEO" tax (Prop D) failed with 47.2% yes. No new money.
    See the data: What each measure did and locks
    1. 2004renewed 2014

      Public Education Enrichment Fund, PEEF (Prop H)

      What it did

      Set aside City General Fund money for public education. A November 2014 measure, also named Prop C, renewed it through June 30, 2041.

      What it locks

      $161.4M in FY26-27. Two-thirds, $107.6M, goes to SFUSD and one-third to preschool. Half of SFUSD's share must go to sports, libraries, arts, and music.

    2. 2008

      Quality Teacher and Education Act, QTEA (Prop A)

      What it did

      Added a parcel tax for teacher pay, $198 per parcel at passage. SFUSD levies it.

      What it locks

      The tax runs through June 30, 2028. After that, voters must renew it.

    3. June 2018replaced 2020

      Educator parcel tax (Prop G), replaced by Fair Wages for Educators (Prop J)

      What it did

      Prop G set a $298 parcel tax for educator pay and was tied up in court. Voters then passed Prop J in November 2020 at $288 per parcel. SFUSD received $123.4M of withheld Prop G money in 2021.

      What it locks

      Runs through FY37-38. The City levies it and appropriates $57.3M in FY26-27.

    4. Nov 2018

      Our City, Our Home (Prop C)

      What it did

      Added a tax on business gross receipts for homelessness services. At least 50% goes to permanent housing and at least 25% to mental health.

      What it locks

      Homelessness uses only. The Board of Supervisors can amend it by a 2/3 vote, and only for its purposes. Any other use needs voters.

    5. Nov 2022

      Student Success Fund (Prop G)

      What it did

      Set aside City money for academics and wellness at school sites, run by DCYF, the City's children's department. About 78% voted yes.

      What it locks

      The City budgets $45.0M in FY26-27 and $60.0M in FY27-28. The Controller treats the FY25-26 freeze as moving the schedule back a year. The Charter text lists $60M for FY26-27 and doesn't say so expressly, so this is a question for the City Attorney. When the March report shows a deficit above a threshold (about $220M), the City may freeze it or cut it, never below $35M. It cannot replace other required school money.

    6. March 2024

      Treatment requirement for county cash aid (Prop F)

      What it did

      People on county cash aid who have a substance use disorder must engage in treatment to keep their aid, with some exemptions, such as SSIP clients. It took effect January 1, 2025.

      What it locks

      An eligibility rule. It sets aside no money.

    7. 2024

      State Prop 1, the Behavioral Health Services Act

      What it did

      Changed how counties spend State behavioral health money from July 1, 2026, with more for housing and full-service care teams and fewer dollars for prevention.

      What it locks

      State rules split county money about 30% to housing, 35% to full-service care, and 35% to other services. San Francisco's adopted budget is $94.6M in FY26-27. DPH's April plan estimated $92.0M, including carryover and Medi-Cal revenue. The Department of Public Health plans to end $973K of school wellness programs on June 30, 2027.

    8. June 2026

      "Overpaid CEO" tax (Prop D), which failed

      What it did

      It would have raised an estimated $250M to $300M a year for the General Fund from 2028. It got 47.2% yes.

      What it locks

      Nothing. It adds no new money, so the City's projected deficits stand.

    Sources

    SFUSD PEEF page; Charter text on PEEF and the Student Success Fund (PDF p. 3: PEEF through FY40-41; p. 8: preschool share; pp. 21-23: Student Success Fund schedule, freeze rule, $35M floor, no replacing other school money); Controller Revenue Letter, June 2026 (Exhibit 4-1, p. 33: $3.09B and PEEF; p. 38: Student Success Fund, about $220M threshold, $57.3M, and who levies each parcel tax; p. 8: Prop D, $250M to $300M a year from FY28-29); Ordinance 140-26, the final budget (PDF p. 67: Student Success Fund; p. 116: Fair Wages $57.3M; p. 181: BHSA $94.6M); SPUR on 2008 Prop A (secondary; $198, levied from July 1, 2008, for 20 years); SPUR on 2020 Prop J (secondary; Prop J at $288, ending June 30, 2038); SFUSD on the $123M (Prop G at $298, $123.4M released in 2021, Prop J through FY37-38); Controller on Our City, Our Home; Ordinance 159-26, p. 2 (two-thirds vote, only to further Prop C's purposes); HSH Budget FY26-28; SF Elections, Nov 2022 final summary report, PDF p. 16 (Measure G, the Student Success Fund: 77.78% yes); City on the Treatment Pathway; DPH deck on the BHSA plan, Health Commission, Apr 2026; Mission Local, June 2026 (press report; election-night count); SF Elections final results; CAAP Eligibility Handbook, Sec. 95-6 (Prop F, the rule from January 1, 2025, SSIP clients exempt); Business and Tax Regulations Code Sec. 2811 (two-thirds vote, only to further Prop C's findings and intent); Charter Sec. 14.101 (voter initiatives change only by voters, unless they say otherwise); SFUSD 2026-27 budget, PDF p. 160 (QTEA expires in 2028; on the ballot for renewal). Legal readings are from public documents and are not legal advice.

    6 of 14 · Our schools

    Reading is at 53.5% in 2026, down from 56.5% in 2019, and poverty predicts most of the gap between schools (est.).

    Each 10 points of low-income share goes with about 8 points lower reading (est.)

    One schoolLabeled schoolThe school picked belowTrend line
    Show data table

    At the typical school over 75% low-income, about 1 in 5 students meets the reading standard; across all their tested students, about 1 in 4 (est.). At the typical school under 40% low-income, about 7 in 10 do.

    • Chronic absence at district-run schools nearly doubled, from 14.2% in 2018-19 to 27.8% in 2025-26.
    • In fall 2024, 1 in 4 SFUSD high schoolers felt sad or hopeless for two weeks or more, down from 35% in fall 2022 (student data).

    Low-income is the State's "socioeconomically disadvantaged" group. Chronically absent means missing at least 10% of school days. This pattern doesn't control for English learners, special education, or selective admission.

    Look up a school: students, seats, absence, and scores

    Look up a school

    What these numbers mean

    Enrollment, low-income share: State census day 2025-26. Scores: share meeting the standard, all tested grades (high schools test grade 11). Capacity: SFUSD's 2022-23 program capacity, which is a seat count set centrally and differs from building capacity. Teachers: full-time equivalents, 2025-26. Ed-Data's 2025-26 staff data show some breaks, so treat this as approximate. "Not reported" means the source left it blank or held it back because fewer than 11 students were tested. Supervisor district is where the school stands, from City address records on the 2022 district map. Families can apply to schools across the city, so students may live in other districts. The 2026-27 plan comes from SFUSD's 2026-27 school allocation files and is subject to change. Staff at average cost plus listed dollars is our estimate. Mission Bay Elementary, in District 6, opened in August 2026 and has no 2025-26 data yet (State school directory; SFUSD school page).

    Who gets paid at SFUSD. Staff get about 77% of the general fund. The top ten vendors got 44% of the $302M paid to outside vendors for daily operations, FY25-26 (est.).

    Operating vendors, $302M paid, FY25-26 (est.)

    The top three are Zum (buses), The Speech Pathology Group (special education services), and Revolution Foods (school meals).

    More than half of vendor money buys four things students use directly: special education services, buses, after-school programs, and meals (est.).

    Limits. Cash, not accrual, and unaudited. Excludes payroll, pensions, and bond debt. OCR name errors are possible. The classification is ours (est.). A big payment does not mean a big margin. Many of these services are required by student plans or funded by restricted State money. Not comparable with the homelessness block.

    Sources

    SFUSD warrant register, June 2026 (one of 12 Board items, all listed under See the data); SFUSD FY25-26 Unaudited Actuals, Form 01, PDF pp. 16-18 ($1,005.7M of $1,310.9M), and Form PCR, PDF pp. 157-158 (transportation coded to special education); SFUSD 2026-27 Recommended Budget, PDF p. 168 (health care); SFUSD warrant register, June 2026, PDF p. 74; The Speech Pathology Group, IEP services, Jun 17, 2025; Bay Area Community Resources, ExCEL contract, Dec 16, 2025.

    See the data: Who gets paid at SFUSD

    SFUSD began posting a monthly check register to Board agendas in October 2025, starting with July 2025 checks. We found one earlier register, for July 1 to August 24, 2023. Of the $643.4M in checks in FY25-26, $301.5M went to operating vendors, $181.1M to construction and capital funds, $103.3M to charter schools, and $57.5M to everything else (est.). This is not total spending. All funds came to about $1.74B (est.), mostly staff, and payroll is not in the register.

    SFUSD now posts a monthly register of vendor checks in its Board packet. We read all 12 months of FY25-26 and matched each month to the Board's total. The largest single payee is Zum, the bus contractor, at $46.6M. SFUSD's accounts code most transportation cost to special education, about 84% (est.).

    Most SFUSD money pays staff: $1.01B of $1.31B in the general fund (FY25-26 actual, unaudited). Teachers are $354M, benefits $318M. CalSTRS is $130M and health care $79M. Health care is budgeted at $114M for FY26-27.

    Limits. Checks are cash, not accrual, and FY25-26 numbers are unaudited. The register excludes payroll, CalSTRS and SFERS pension payments, main medical plan payments, and bond debt. It does include dental and some other benefit payments. It was read by text recognition (OCR): dollar totals match the Board's totals exactly, but names may have small errors. Service types and the operating vs. capital split are our classification (est.). A big payment does not mean a big margin or overcharging. We did not test prices. Many of these services are required by student plans or funded by restricted State money. This block is not comparable with the homelessness block, which counts nonprofit payments only. This one counts all outside vendors.

    The vendor list has the 367 organizations paid $100,000 or more, covering $611.0M of the $643.4M. Names follow the register. We merged most spelling variants and OCR errors, so a few organizations still appear on two lines. We grouped some related company names, such as AT&T and AT&T Mobility, and we grouped schools under one charter family. We did not verify legal entities. Payments to individuals and families from operating funds total $11.0M (est.), and we do not list them. Another $0.4M to individuals sits in the self-insurance and capital lines. Union and benefit plan rows are mostly money withheld from pay or owed for staff benefits, not fees for a service (est.).

    All $643M in checks, and what the $302M buys (FY25-26, paid, classified by us, est.)

    All $643M in checks

    What the $302M buys

    Bond money pays for school facilities projects, including technology and kitchens. Voters restricted it to listed projects, so it cannot pay for daily operations. Most of the $181M to builders, architects, and other capital work is bond money ($162.5M from the Building Fund, est.). Another $103M passed through to charter schools.

    Pie labels are rounded, so they can add to a little more or less than the total. Sources: SFUSD warrant registers (12 Board items, listed below); 2016 bond audit, FY24-25; 2024 bond audit, FY24-25.

    Top 25 operating vendors, FY25-26 (paid; funds 01, 05, 11, 12, 13)

    Top 3 are $79.36M, 26.3%. Next 7 are $54.25M, 18.0%. The other payees, about 2,000, are $167.90M, 55.7% (est.). The Board contract list column is the prior year's authority to spend, not payments. It cannot show whether FY25-26 payments were within their authority. We found no FY25-26 list, and new FY25-26 contracts, such as the InquirEd adoption, are not on it. "Not listed" only means the name is not on that list. Utilities and bond construction vendors do not appear on it.

    "What they do" comes from Board items and contract lists, and sources are linked where one item covers it. Some linked items are from other years. IEP means a student's special education plan. Sources: SFUSD warrant registers (listed below); SFUSD contract list, FY24-25.

    All 367 organizations paid $100,000 or more, FY25-26, and how each month matched the Board's total

    The 367 cover $611.0M of the $643.4M. Types are our labels (est.). A row's type follows the largest part of what it was paid. Each month's total matched the Board's cover total to the penny.

    Show all 367 organizations
    Show each month against the Board's total

    Sources: SFUSD warrant registers (listed below). Names follow the register, read by text recognition, so they may have small errors.

    Construction and capital funds, FY25-26 (paid): the 13 largest payees

    Funds 14, 21, 25, 30, 35, 40, 49. Top 3 are $58.40M, next 7 are $60.88M, the rest are $61.86M (est.). Most is Proposition A bond money. The FY24-25 audits of the 2016 and 2024 bonds found "no findings or recommendations." They cover FY24-25, not these FY25-26 checks. Commitments on unfinished projects were $167.5M at June 30, 2025.

    Amounts are net of retention held back (object 9515). Payments to banks coded to construction objects look like retention paid to escrow (est.). Sources: SFUSD warrant registers (listed below); 2016 bond audit, FY24-25, PDF p. 25; 2024 bond audit, FY24-25, PDF p. 22; SFUSD FY24-25 audit, PDF p. 65.

    Charter schools, $103.3M, FY25-26 (paid). This is not a vendor relationship

    Almost all of this is public school funding that follows students, not a vendor relationship. Lines: $64.61M from Fund 77 (agency pass-through), $31.37M coded 01-8096 (in lieu of property taxes), $3.74M coded 01-7211, $3.54M other, with purposes not stated in the register (est.). Five Keys enrolls 3,487 of the 7,388 charter students, including adults in county jails. SFUSD's accounts show the in-lieu transfer as $32.02M, recorded as negative revenue, not spending.

    Sources: SFUSD warrant registers (listed below); SFUSD FY25-26 Unaudited Actuals, PDF p. 13.

    Staff are 76.7% of the general fund, FY25-26 actual (est.; $M)

    • The "PERS" line. The audit says classified employees are SFERS members, not CalPERS. The form's "PERS" line was zero through FY24-25, while the OASDI and alternative line was higher. The coding appears to have moved in FY25-26 with the new system (est.).
    • Pension payments, audited. District CalSTRS contributions were $91.06M in FY24-25, and the State paid about $41.29M on SFUSD's behalf. SFERS contributions were $18.42M. So the CalSTRS line includes money SFUSD did not raise itself (est.).
    • Health care is budgeted up 43.9% from FY25-26 actual to FY26-27, from $79.36M to $114.18M. SFUSD says parcel taxes will cover the increase if needed.
    • Administrators (1300 plus 2300) were $63.12M, 4.8% of the general fund in FY25-26 (est.).
    • Other shares. All funds, 62.6%. Operating funds only, without capital and debt, 76.0% (est.). Unrestricted general fund, 87% (FCMAT, from second interim), against a 2024-25 statewide average of 86%.

    Sources: FY24-25 Unaudited Actuals, PDF pp. 12-15; FY25-26 Unaudited Actuals, PDF pp. 16-18 (FY26-27 is column F, matching the budget); SFUSD FY24-25 audit, PDF pp. 55, 56, 60; SFUSD 2026-27 Recommended Budget, PDF p. 168; FCMAT, Aug 2026, PDF p. 24.

    Services and contracts by object: $239.3M actual in FY25-26, against a $211.3M budget for FY26-27 ($M)

    The $211.3M budget for services and contracts breaks down to $105.5M professional and operating services, $74.8M subagreements, $17.8M operations, $8.4M rentals, and $4.8M other. Actual spending has come in above the adopted budget for three years. In FY25-26 it was $239.3M. Budgets change during the year: SFUSD says it captured $69.7M from underspent non-labor accounts at FY25-26 third interim. It plans $29M less in contracts. Treat $211.3M as a plan.

    General fund, Form 01, 5000-5999. Insurance (5400) is zero in the general fund. Insurance sits in the self-insurance fund. Adopted figures are column F of the prior year's report. Sources: FY20-21, PDF p. 17; FY21-22, PDF p. 15, scanned; FY22-23, PDF p. 9; FY23-24, PDF p. 14; FY24-25, PDF p. 13; FY25-26, PDF p. 17; SFUSD 2026-27 Recommended Budget, PDF p. 5.

    By bargaining unit, FY25-26

    Public disclosure forms required by AB 1200, FY25-26. They are the only public unit-level pay totals we found. Four of the five forms are scanned and were read by OCR. Not all benefit costs are stated, and health plan costs are mostly blank. FTE means full-time equivalent. Source links are in the table.

    School sites vs. central office. No one publishes a clean current split

    • Definitions drive the answer. The City Budget Analyst counted districtwide instructional supervision ($104.5M) and plant ($65.9M) as central. SACS, the State's school accounting format, counts neither as general administration. SFUSD's "centrally budgeted" includes transportation $33.8M and custodians $31.6M. Those are delivered at schools.
    • The Budget Analyst's figure is old. It predates 205 central positions cut for FY25-26. We could not repeat its method for later years without the State's detailed files.
    • Contract lists understate school-level services. After-school, buses, meals, and special education appear under central orgs.
    • State and auditors do not give a split. FCMAT's 2024 and 2026 reviews and the audits do not report a school vs. central share.
    • School plans give part of the picture. SFUSD's 2026-27 school allocation files list staff and some dollars for each of 102 schools, not totals. At average cost, they come to about $547M, or about $11,400 per projected student (est.). That leaves out special education teachers and aides, other paraeducators, nurses, custodians, food, buses, substitutes other than dedicated core subs, and central offices, so this isn't the school share of spending.

    All funds, FY25-26 actual: $1,742.81M with no double count (est.)

    The FY26-27 budget is shown only to warn against using it: capital and debt are placeholders there (Building Fund $11.47M vs. $152.49M actual).

    Double counting (est.): transfers of $137.11M are not in these totals, indirect charges between funds net to zero, and bond spending and debt service count the same capital dollars at different times, so do not add them across years. Outside these totals: charter pass-through of about $64.6M through Fund 77, and in-lieu property tax to charters of $32.02M, recorded as negative revenue. Sources: SFUSD FY25-26 Unaudited Actuals, PDF pp. 10, 25, 32, 39, 46, 53, 64, 71, 78, 85, 92, 99, 106-107, 118-119, and 156; SFCOE FY25-26 Unaudited Actuals, PDF p. 10.

    The EMPower payroll system

    Board approvals for EMPower are contract authority, and they overlap by year, so they can't be added into a total cost. The Board adopted Frontline as the replacement on Mar 12, 2024, and it went live with FY25-26.

    Sources: Infosys, through the third amendment; Infosys, fourth amendment; Alvarez and Marsal, Sep 2022; Alvarez and Marsal, Dec 2022; SAP, through Apr 2023; SAP, later approval; SAP, latest approval; Board adopts Frontline; FCMAT, Aug 2026, PDF p. 9.

    Limits

    • Checks are cash, not accrual. Some FY25-26 checks paid earlier bills, about $54.8M coded to object 9510 (est.), and some FY25-26 bills were paid later.
    • FY25-26 numbers are unaudited.
    • The register covers what the Board calls vendor warrants, from one bank account. It includes charter pass-through, dental and some other benefit payments, and payments to individuals. It excludes payroll, CalSTRS and SFERS pension payments, main medical plan payments, student activity accounts, and bond debt.
    • Registers are image PDFs read by OCR. Dollar totals match the Board's totals exactly. Names may have small errors.
    • Service types and the operating vs. capital split are our classification (est.).
    • A big payment does not mean a big margin or overcharging. We did not test prices. Many of these services are required by student plans or funded by restricted State money.
    • This block is not comparable with the homelessness block, which counts nonprofit payments only. This one counts all outside vendors.
    Sources

    SFUSD warrant registers, FY25-26 (12 Board items): July 2025; August 2025; September 2025; October 2025; November 2025; December 2025; January 2026; February 2026; March 2026; April 2026; May 2026; June 2026. SFUSD FY25-26 Unaudited Actuals, Form 01; SFUSD 2026-27 Recommended Budget, PDF p. 5; SFUSD contract list, FY24-25; Zum Seventh Amendment; FCMAT Fiscal Health Risk Analysis, Aug 2026; SFUSD FY24-25 audit; 2016 bond audit, FY24-25 (bond money limited to listed projects); SFUSD warrant register, Jul 1 to Aug 24, 2023, Board item Sep 12, 2023; InquirED agreement, $3,679,186.99, Apr 28, 2026; 2024 bond audit, FY24-25, PDF p. 22; SFUSD Central Office Reduction Plan, Apr 22, 2025, PDF p. 4; The Speech Pathology Group, IEP services, Jun 17, 2025; YMCA ExCEL resolution, $4,551,261, Dec 16, 2025; BACR ExCEL resolution, $2,929,905, Dec 16, 2025; Amergis agreement, Jun 24, 2025; Stepping Stones Group agreement, Jul 29, 2025; Ro Health, nonpublic agency agreements, Oct 27, 2026; Mission Graduates ExCEL resolution, $1,261,836, Dec 16, 2025; Dell Chromebook purchase order through Arey Jones, $1,867,112.15, Mar 10, 2026; RDNC ExCEL resolution, $1,285,540, Dec 16, 2025; North Valley School, a division of Victor Treatment Center, master contract, Sep 9, 2025; Progressus Therapy agreement, Jul 29, 2025; Robert Half-Protiviti contract, $2,000,000, Jun 17, 2025; RCM agreement, Jun 17, 2025; Jamestown ExCEL resolution, $1,106,123, Dec 16, 2025; CYC ExCEL resolution, $1,333,924, Dec 16, 2025; Buhler Commercial, contract 6264 modification 4, Mar 24, 2026; Angotti & Reilly, contract 6276 modification 3, Apr 14, 2026; Rodan Builders, contract 6292, Apr 22, 2025; Mar Con Builders, contract 6298 modification 9, May 26, 2026; Sierra School Equipment, purchase agreement 6099, Aug 13, 2024; Jackson Liles Architecture, contract 6069, Aug 13, 2024; Best Contracting Services, contract 6273, Apr 8, 2025; AECOM, contract 6307, May 13, 2025.

    About 1 in 3 SF K-12 students attend private school, three times the California share (est.). State law does not require private schools to give the State test.

    Families choose or apply to charters and private schools, so no comparison here shows that one kind of school works better. Differences can reflect who attends.

    Private share of K-12 students, 2020-2024 (Census survey estimates, est.): 31.6% (+/-1.3) in San Francisco, 10.0% (+/-0.1) in California, with 90% margins.

    City choices about SFUSD reach about two of every three SF K-12 students (est.).

    Among all low-income students, reading is close at charters and district-run schools (37.6% vs 38.5%), and math is lower at charters (25.6% vs 32.2%) (est.). The mix of students differs: among low-income Hispanic students, charters are higher in reading (36.7% vs 19.9%) and math (25.1% vs 12.6%) (est.).

    Sources

    Census ACS, table B14002, 2020-2024 (raw data file); State private school data, 2024-25; State enrollment, 2024-25; Ed. Code 60640; Ed. Code 60603; State test results, 2026 list of schools; State test results, SF file, 2026; Census PUMS data dictionary, 2020-2024 (SCH value 3); SFUSD Student Assignment Policy; Census table B14002, San Francisco (table view); SFUSD 2023 Facilities Master Plan, PDF p. 19.

    See the data: Private and charter schools

    Families choose or apply to charters and private schools. SFUSD families also rank district schools, and some district schools select students. Differences can reflect who attends. No comparison here shows that one kind of school works better.

    Private share of K-12 students, 2020-2024 (Census survey estimates, est.)

    Show data table

    San Francisco's share is the highest among the 50 largest U.S. cities (est.). Milwaukee is second at 25.2% (+/-1.2). The chart shows selected places. Survey estimates with 90% margins. "Private" includes home school. The Census counts where children live, not where they go to school.

    California requires public districts, county offices, State special schools, and direct-funded charters to give the State test. Private schools are not on that list. Missing data is not a finding either way.

    Sources: SFUSD 2023 Facilities Master Plan, PDF p. 19; SFUSD Student Assignment Policy.

    About 1 in 3 SF K-12 students attend private school, three times the California share (est.)

    City choices about SFUSD reach about two of every three SF K-12 students (est.).

    The Census counts about 25,000 SF residents in private K-12 school (24,959, +/-1,198), or 31.6% (+/-1.3). That is the highest share among the 50 largest U.S. cities (est.). Milwaukee is second at 25.2% (+/-1.2), and this table shows selected places. State counts of students at schools in SF give about 31%, not counting Five Keys (est.).

    Limits: Survey estimates with margins. "Private" includes home school. The Census counts where children live, not where they go to school. Sources: Census ACS, table B14002, 2020-2024 (raw data file); State private school data, 2024-25; State enrollment, 2024-25; Census PUMS data dictionary, 2020-2024 (SCH value 3); Census table B14002, San Francisco (table view).

    Private school enrollment in SF has stayed near 24,000 to 25,000 for 20 years (est.)

    Since 2019-20, private schools in SF did not add students overall once we fill gaps in filings, while district-run schools lost about 3,900 (est.).

    SF had 107 private schools with 24,218 students in 2024-25. Fewer are in the early grades and more are in high school than in 2005-06. From 2019-20 to 2024-25, private schools changed by about -400 (est.), and district-run schools went from 52,778 to 48,902 (est.). These numbers don't show where families went.

    Limits: Schools report their own numbers, only schools with six or more students are listed, and some skip a year. We carried forward the last filing for schools that filed again later. The 2025-26 file leaves out 17 schools that filed the year before, including two of the largest, so we stop at 2024-25. Sources: State private school data; FCMAT, Apr 2024, PDF p. 23; State enrollment, 2024-25.

    None of SF's 107 private schools appear in the State's test results. The law does not require them to take part

    We found no public results by school on a shared test, so we do not compare private and public schools on test scores.

    California requires public districts, county offices, State special schools, and direct-funded charters to give the State test. Private schools are not on that list. The State says it has no authority to evaluate or approve private schools. Missing data is not a finding either way. The national studies we read find that raw private school advantages shrink a lot or disappear once student and family background are taken into account, and results vary by grade and subject.

    Limits: The research is national, not on SF, and none of it shows one kind of school works better for a given child. Schools may post their own scores or college lists. We don't use them, because they aren't comparable. Sources: Ed. Code 60640; Ed. Code 60603; State, CalEdFacts on private schools; State test results, 2026 list of schools; NCES, 2006; Lubienski & Lubienski, 2006; Pianta & Ansari, 2018; NAEP data, grade 8 reading by school type, 2024; NAEP urban districts.

    63% (+/-5) of SF K-12 students in households earning $500,000 or more attend private school, against 15.5% (+/-2.6) in households under $100,000 (est.)

    Who attends differs sharply by family income, which is one reason raw comparisons with public schools mislead.

    Published tuition at 12 schools we checked runs from about $5,000 at a Catholic K-8 to about $68,000 at an independent high school. Catholic high schools we checked are near $30,000, and most independent K-8 schools we checked are near $50,000. Six independent schools say 23% to 40% of students get aid. Private students are 41% non-Hispanic White (+/-3.1) and 3% non-Hispanic Black (+/-1.4), against 18% (+/-1.3) and 7% (+/-0.9) of public students (est.).

    Limits: Survey estimates, 2020-2024. "Private" includes home school. Tuition is what schools post, for 12 of about 100 schools, and it is not what every family pays. We found no public source that gives low-income shares by private school. Sources: Census ACS microdata, 2020-2024; school tuition pages: Mission Dolores Academy; San Francisco Christian School; Kittredge School; Archbishop Riordan; Sacred Heart Cathedral; SF High School of the Arts; Chinese American International School; Live Oak School; San Francisco Friends School; Cathedral School for Boys; Drew School; San Francisco University High School; University High aid page (23% of students, 2026-27); Census PUMS data dictionary, 2020-2024 (SCH value 3).

    Per 100 12th graders, UC admitted about 49 from SF private schools, 38 from district-run schools, and 35 from charters, and about 17, 21, and 11 enrolled (est.)

    This is the one result we found reported the same way for all three, and it shows who applies as much as how schools do. Of those who applied, UC admitted 72% from private schools, 81% from district-run schools, and 78% from charters (est.).

    Fall 2025. Grade 12 counts are from 2024-25. Limits: UC is one college system. Families choosing other colleges don't appear, and applying and enrolling reflect preference as well as readiness. It does not show what a school added. UC blanks small counts, so totals run slightly low. Sources: UC admissions by source school; State enrollment, 2024-25; State private school data, 2024-25.

    Among all low-income students, reading is close at charters and district-run schools (37.6% vs 38.5%), and math is lower at charters (25.6% vs 32.2%). Among low-income Hispanic and Black students, charters are higher (est.)

    Results vary widely from charter to charter. Some run well above similar district schools, and some well below (est.). Who is tested also differs: about 70% of low-income students tested at charters are Hispanic, against 41% at district-run schools, where 36% are Asian (est.).

    For all students, not only low-income students: compared with district schools that have the same grades and a similar low-income share, about half of charter grade groups were above and half below in reading in 2026, with 7 above and 9 below in math (est.). Charters here means every charter in SF except Five Keys, which serves adults, including people in county jails. Five Keys tests few students, so it is not in these tables.

    Limits: Families choose or apply to charters, so these gaps don't show that one kind of school works better. The charter group is small: about 1,500 low-income students tested, in 12 schools. The schools in it changed from 2019 to 2026, and one school accounts for the 2019 reading edge (est.). Sources: State test results, SF files, 2019; 2022; 2026; Five Keys mission page (its own page, undated).

    Chronic absence was 25.8% at SFUSD-authorized charters and 27.3% at district-run schools in 2025-26, not counting alternative-status schools

    Five Keys and other alternative-status charters explain why SFUSD's rate jumps from 27.8% to 32.7% once charters are counted.

    Alternative-status schools are Five Keys and Life Learning among charters, and Downtown and Ida B. Wells among district-run schools. Charters ran higher in each earlier year. Counting all schools except alternative-status ones, SFUSD's 2025-26 rate is 27.2%.

    Limits: 2025-26 included the four-day February 2026 strike by SFUSD's teachers, and we could not confirm how those days were counted. Don't read one year as a trend. Families choose charters, so gaps don't show cause. Sources: State chronic absence, 2025-26; 2022-23; 2023-24; 2024-25; The Frisc, Feb 2026 (press).

    SF had 14 charter schools in 2025-26; outside Five Keys they enroll 3,901 students, down 15% from 2019-20 (est.)

    Outside Five Keys, charters serve about 7% of SF's public K-12 students (est.), and none has opened since 2019.

    Twelve are under SFUSD's code, and two were under the State Board in 2025-26: The New School and KIPP Bayview Elementary. KIPP Bayview Elementary has since moved to SFUSD. Five Keys Charter closed in June 2026. None is authorized by the County Office. The Board rejected a new charter in August 2025.

    Limits: SFUSD counts 11 charters, and counting rules differ. Ed-Data shows 7 Five Keys students in 2009-10 and 642 the next year; we could not confirm why. Five Keys works in 14 counties, and its students aren't split by county. We could not count SF residents who attend charters elsewhere. Sources: State school directory, Oct 2026; Ed-Data charter enrollment; State enrollment, 2025-26; SFUSD charter oversight page; KIPP Bayview Elementary staff report; The Frisc (press); Dragon Gate staff report, Aug 2025; Five Keys mission page (its own page, undated); Ed-Data charter enrollment, 2009-10 to 2018-19; Ed-Data Census Day enrollment, 2009-10 to 2018-19; SFUSD Facts at a Glance, 2025-26; SFUSD Board minutes, Aug 26, 2025, PDF p. 2; SFUSD Board minutes, Oct 28, 2025, PDF p. 2; State school directory: KIPP Bayview Elementary.

    SFUSD budgets $30.5M in property tax money for charters in FY26-27, after $32.0M in FY25-26, and it must offer them space

    State formula money follows each student to a charter. SFUSD can charge its actual oversight costs, up to 1% of a charter's State formula revenue, or 3% if it provides space nearly rent free.

    SFUSD staff estimated in 2025 that one new 396-student charter could cost the district more than $5 million a year in lost funding, or about $12,600 a student (est.). State law, amended in 2026, requires SFUSD to visit each charter site it authorizes yearly, check its reports, and monitor its finances. It must offer space on request for students who live in the district. It may refuse a request based on fewer than 80 students in average attendance. SFUSD's community oversight committee stopped meeting in November 2023.

    Limits: Gateway is one example. Its audit is posted with its renewal petition. We did not compare other charters' budgets, so we can't say it is typical. The table leaves out other revenue, such as $3.9M in grants and donations. We could not find SFUSD's total oversight fees, Prop 39 charges, or parcel tax paid to all charters. The $5 million is a staff estimate for a school that was never opened. The report does not subtract costs SFUSD would avoid. Sources: SFUSD 2026-27 budget, Form 01, PDF p. 20; Ed. Code 47635; 47613; 47614; 47604.32; Gateway petition and audit, PDF pp. 46, 262, 270, 274; Dragon Gate staff report, PDF pp. 2, 22; SFUSD oversight page; SFUSD FY25-26 Unaudited Actuals, Form 01, PDF p. 13.

    Parent and teacher voices: coming soon.
    See the data: Scores, absence, and the exceptions

    Share of SFUSD students meeting the State standard

    19%Median reading at schools over 75% low-income, 2026 tests (est.). Math is 13%.
    73%Median reading at schools under 40% low-income, 2026 tests (est.). Math is 65%.

    Chronic absence at district-run schools nearly doubled, from 14.2% in 2018-19 to 27.8% in 2025-26

    Counting SFUSD-authorized charters, the rates are 18.8%, 28.8%, and 32.7%. Five Keys and other alternative-status charters account for the jump from 27.8% to 32.7%. Without alternative-status schools, the State's 2025-26 rate is 27.2%. Chronic absence rose at 82 of 101 district-run schools in the last year and at 98 since 2018-19, and 30 schools are at 40% or higher. 2025-26 included the four-day February 2026 teacher strike; we could not confirm how those days were counted.

    The exceptions

    Four schools over 70% low-income had at least half their students meeting the reading standard in 2026: John Yehall Chin (83%), Marina Middle (55%), Gordon J. Lau (55%), and Jean Parker (54%). Among the 99 schools charted, one under 40% low-income scored below 50%: Harvey Milk Civil Rights Elementary (46%). The County Office's special education program (30.8% low-income) is not charted; its 2026 reading score is not reported (6 tested).

    What the words mean. Reading is the State's English language arts test. Low-income is the State's "socioeconomically disadvantaged" group, which counts students who qualify for free or reduced-price meals and some other groups, such as students whose parents did not finish high school, as a share of 2025-26 enrollment. Chronically absent means missing at least 10% of school days. Absence figures are for district-run schools. This is a pattern across 99 district-run K-12 schools; Lee Newcomer and Mission Education Center have no 2026 reading score. It does not control for English learners, special education, or selective admission at Lowell and Ruth Asawa School of the Arts.

    See the data: Student mental health and drug use

    In fall 2024, 1 in 4 SFUSD high schoolers felt sad or hopeless for two weeks or more, down from 35% in fall 2022

    24.3%High schoolers who felt sad or hopeless almost every day for 2 weeks or more in the past year (35.3% in fall 2022; 24.9% in 2015)
    8.8%High schoolers who seriously considered suicide in the past year (15.4% in fall 2022); 3.7% attempted
    16.6%Middle schoolers who have ever seriously considered suicide (20.5% in fall 2022)
    39.7%High schoolers who went to their school's Wellness Center at least once in the past year
    • Some students report much more. Among lesbian, gay, or bisexual students, 51.8% of high schoolers felt sad or hopeless and 28.8% seriously considered suicide, and 44.4% of middle schoolers have ever considered it. Among high schoolers, 32.6% of girls and 16.1% of boys felt sad or hopeless.
    • Drugs at school fell. 8.9% of high schoolers were offered, sold, or given an illegal drug at school in the past year, down from 13.3% in fall 2022 and 23.8% in 2015. 7.1% of high schoolers have ever taken prescription medicine without a prescription. 5.7% of middle schoolers have ever misused prescription pain medicine.
    • Young people are a small share of overdose deaths. 11 people aged 15 to 24 died of accidental overdoses in San Francisco in 2025, 2% of all 625. None were under 15.
    • Staffing. SFUSD had 99 school-site social workers for 50,028 students in 2025-26, about 1 per 500 (est.). The School Social Work Association of America recommends no more than 250 students per social worker. A proposal to cut 45 of the 99 was reversed in December 2025. A proposal to end middle school health and wellness programs has no reported outcome (Cuts).
    • City money. DPH plans to end $973K of school wellness programs on June 30, 2027, and no opioid settlement money has been found going to schools (The ask, Parts 1 and 2).
    • What works. In 43 controlled trials of mental health services that school staff gave elementary students, problems fell (effect size 0.39), most for students already showing problems (0.76) and least for programs for all students (0.29). Programs that combine social skills and resistance training had small but consistent protective effects against drug use in a 2014 review of 51 trials; the one trial of an information-only program found no effect on drug use.

    SFUSD's Youth Risk Behavior Survey is self-reported and weighted to all regular and alternative SFUSD students: 3,605 high schoolers (61% of those sampled) and 2,478 middle schoolers (62%) in fall 2024. SFUSD labels the fall 2024 round "2025" and the fall 2022 round "2023". It reports middle and high school totals only, not schools. The 43 trials covered elementary students, not teens. These figures show need and use of services, not that more money would lower the rates.

    Sources

    SFUSD Youth Risk Behavior Survey slides: high school, fall 2024; middle school, fall 2024; high school, fall 2022; middle school, fall 2022; SFUSD's index of survey reports; SF Medical Examiner, accidental overdose report, 2025, p. 3; School Social Work Association of America, ratio resolution; SF Standard, June 26, 2026 (press; 99 school-site social workers); SFUSD Superintendent message, Dec 18, 2025 (social workers kept); DPH deck on the BHSA plan, Health Commission, Apr 2026, slide 18 ($973K); Sanchez et al., Journal of the American Academy of Child and Adolescent Psychiatry, 2018 (43 trials); Faggiano et al., Cochrane review of school-based drug prevention, 2014 (51 trials).

    See the data: School by school

    SFUSD runs 122 schools; most are elementary

    SFUSD-run schools by level, 2025-26

    SFUSD counts another 11 charter schools separately. Source: SFUSD Facts at a Glance, 2025-26.

    Every school: 106 rows, 101 district-run K-12 schools with 2025-26 data, 4 County Office schools, and 1 closed school, with scores, absence, capacity, and SFUSD's budget plan

    Mission Bay Elementary, in Supervisor District 6, opened in August 2026, so it has no 2025-26 data and isn't listed here. The school lookup has its 2026-27 plan. Source: State school directory.

    The 2026-27 plan columns come from SFUSD's 2026-27 school allocation files, revised March and April 2026 and subject to change. Listed dollars are on top of staff positions. Per student is our estimate: staff at average cost plus listed dollars, per projected student, blank for schools under 50 projected students. It leaves out special education teachers and aides, other paraeducators, nurses, custodians, food, buses, substitutes other than dedicated core subs, and central offices, so it isn't a school's full cost. The Academy has no 2026-27 plan.

    Show SF charters (14)

    Five Keys programs serve adults, including people in county jails (3,487 of 7,388 charter students); compare them with care. Scores at very small groups swing widely: Life Learning tested 14 students in 2025, and Five Keys Independence tested 11 in reading. A blank means our files have no value for that school.

    Select a column heading to sort. Reading and math are the share meeting or exceeding the standard (2025 = spring of 2024-25). Low-income is the State's socioeconomically disadvantaged share of 2025-26 enrollment, which is different from the State's unduplicated count. Blank = suppressed (under 11 tested), not reported, or a homeless count under 11. Capacity is SFUSD's 2022-23 program capacity; % full = 2025-26 enrollment over that capacity (est.). Sources: CDE Census Day 2025-26; CAASPP research files; CDE chronic absence 2025-26; SFUSD 2023 Facilities Master Plan, PDF pp. 76-81 (its column is called Enrollment capacity); Ed-Data teacher FTE (chart 180; this link shows Lowell, and each school has its own chart); CDE school directory; DataSF Analysis Neighborhoods.

    14 schools are under 70% of capacity; 9 also have under 30% meeting the reading standard

    Also under 30% in readingOther

    2025-26 enrollment over 2022-23 program capacity (est.). Across 100 schools: 47,410 students in 52,997 program seats (89.5%). Sources: CDE Census Day; 2023 Facilities Master Plan; CAASPP 2025.

    Limits on these school figures

    • Capacity is program capacity, not building capacity. It is SFUSD's centrally set 2022-23 seat count from the 2023 Facilities Master Plan. A draft Board resolution says SFUSD's 125 campuses have room for 14,000 more students (press report).
    • Staffing sources disagree. Ed-Data's full-time-equivalent teachers attributed to district-run schools rose from 2,274 (2018-19) to 2,566 (2025-26). Its district-wide figure, which also counts charter teachers (306 in 2025-26), fell from about 3,775 to 2,857. Schools plus charters add up to that figure in 2025-26, but about 1,160 were not attributed to any school in 2018-19, which may be a coding change (est.). SFUSD's own count of teachers is 3,387 (2022-23) to 2,947 (2025-26). These are different counts and are not comparable. Ed-Data's 2025-26 staff data show breaks.
    • The 2026 test measure was renamed "Percentage Proficient and Above"; no ETS note confirming unchanged cut scores was found.
    • We found no per-school split of counselors, social workers, or nurses. Ed-Data's school charts show them only as combined pupil services.

    Sources: 2023 Facilities Master Plan; SF Standard, Dec 2025 (press, quoting a draft Board resolution); Ed-Data district teacher FTE; SFUSD Facts at a Glance; ETS research file list.

    Sources

    Five Keys mission page (founded by the SF Sheriff's Department for adults in county jails; now also community centers); Ed-Data teacher FTE by school, chart 180 (we summed the school charts; this link shows Lowell); Ed-Data administrators and pupil services (chart 187; this link shows Lowell).

    See the data: SFUSD budget

    SFUSD balanced FY26-27 after about $114M in cuts, and the gap returns in FY28-29: about $26M unrestricted, with restricted funds running deficits, in a projection that assumes voters renew the QTEA parcel tax before it ends June 30, 2028.

    State formulas and grants set about 70% of SFUSD's revenue (State cash about 46%, est.); salaries and benefits take about 80% of its FY26-27 budget

    Revenue, FY26-27, $1.38B

    General fund spending by type, FY26-27 budget, $1.36B

    General fund spending by function, FY24-25

    Revenue includes the County Office. The $1.26B function total is the full general fund; the five lines in the table below add to $1,250.4M. About $329M of the State formula is local property tax, so State cash is about 46% and all local money about 50% (est., SFUSD general fund). Sources: SFUSD 2026-27 budget, Form 01 (PDF pp. 19-20 revenue; pp. 23-25 spending); CDE Budget Overview; SF Standard, June 2026 (press); Ed-Data function chart.

    The City sends SFUSD about $168M (est.) through Charter set-asides, nearly two-thirds of it PEEF

    Show data table

    Charter set-asides total about $167.9M (est.). PEEF share is $107.6M, two-thirds of $161.4M. Source: SF Controller Revenue Letter, June 9, 2026.

    The Student Success Fund granted SFUSD less than the City appropriated in each year with data

    The Controller treats the FY25-26 freeze as moving the schedule back a year. The Charter text doesn't say so expressly, so this is a question for the City Attorney. Unspent money sits in an $18.1M reserve the Controller's plan does not touch through FY29-30. Sources: Controller, June 2026; Controller March 2026 Update; Res. 483-24; Res. 471-25.

    Salaries and benefits are about 77% to 80% of SFUSD's general fund ($M)

    Services came in above the adopted budget three years running: $215.7M against $201.3M in FY23-24, $236.5M against $190.5M in FY24-25, and $239.3M against $218.6M in FY25-26. The $211.3M is the FY26-27 budget.

    About 80% goes to salaries and benefits in the FY26-27 budget. In FY25-26 actual, it was 76.7%, $1,005.7M of $1,310.9M (est., unaudited). Total includes capital outlay and other outgo not listed. Sources: SFUSD FY25-26 Unaudited Actuals, PDF pp. 16-17 (FY25-26 actual; FY26-27 budget); Ed-Data by object (five lines, FY23-24 and FY24-25); SF Standard, June 2026 (press).

    Enrollment fell from 61,031 in 2019-20 to 55,237 in 2025-26

    Includes SFUSD-authorized charters. Source: CDE DataQuest.

    Homeless students, counting charters, doubled to 3,624 in two years

    50.9% were chronically absent in FY24-25 counting charters (43.5% at district-run schools). SFUSD refined the housing survey on its online emergency card to find more students, which may explain part of the FY24-25 jump (our reading). Sources: Ed-Data counts; Ed-Data absence.

    SFUSD spends 26% more per student than the state average, but less than LA and Oakland, FY24-25

    Show data table

    Current expense of education per ADA (average daily attendance). Source: CDE Current Expense of Education.

    Sources

    SFUSD 2026-27 budget, PDF pp. 8, 131-133 (multi-year projection; QTEA renewal assumed); Form 01, PDF pp. 23-25 (FY26-27 column); SPUR on QTEA (secondary; levied July 1, 2008, for 20 years); Controller, June 2025 Revenue Letter, PDF pp. 29, 34 (FY25-26: $35.0M; deficit $272.3M over $208.2M threshold); Charter text, Sec. 16.131(d), PDF pp. 21-23 (schedule, 3% cap, $35M floor, reserve); Ordinance 140-26, PDF p. 67 (Student Success Fund $45.0M and $60.0M, adopted); Ed-Data spending by function (totals $1,208.9M and $1,262.8M); State chronic absence 2024-25 (43.5% of homeless students at district-run schools); SFUSD SAFEH report, 2025, PDF p. 7 (housing survey on the emergency card refined).

    See the data: SFUSD over 20 years

    SFUSD's real spending per student roughly doubled from $13,393 in FY05-06 to $27,642 in FY24-25 (FY25-26 dollars, est.), while attendance fell 10%. SFUSD's own count of teachers fell from 3,387 (2022-23) to 2,947 (2025-26); sources count teachers differently. Most of the real gain came after the State's new formula began in FY13-14: about 43% from FY13-14 to FY18-19 and about 51% from FY18-19 to FY24-25 (est.).

    Real spending per student rose 106% in 19 years (est.)

    Current expense of education per unit of average daily attendance (ADA): general fund operating spending, excluding capital, debt, and food service. Real = FY25-26 San Francisco dollars, using fiscal-year averages of the SF-area CPI-U (est.). Real spending fell 9% from FY09-10 to FY12-13, after the recession; FY21-22 jumped 30% as attendance fell 11% in one year. Sources: CDE Current Expense of Education; BLS CPI-U, San Francisco area.

    Attendance dropped by about 5,400 in FY21-22 and has not come back

    Census Day enrollment includes SFUSD-authorized charters; FY05-06 to FY08-09 come from Census Bureau fall membership, which matches Ed-Data exactly in FY09-10. Attendance (ADA) excludes charters, so the gap between the lines is not an absence rate. Sources: Ed-Data enrollment; Census school finance survey; CDE ADA.

    Ed-Data's full-time-equivalent teacher count went from 3,348 in FY09-10 to 2,857 in FY25-26

    This is Ed-Data's count, with breaks; FY17-18 (4,622) is an apparent reporting artifact. It is not comparable with SFUSD's own count of teachers (3,387 in 2022-23, 2,947 in 2025-26), so do not read one percentage change across the two. Sources: Ed-Data chart 180; SFUSD Facts at a Glance.

    On Ed-Data's count, students per teacher rose from 16.5 to 21.0 by FY24-25, then 19.3

    Selected years, using Ed-Data's teacher count. Source: Ed-Data chart 192.

    SFUSD runs 122 schools, and the K-12 count has barely moved since FY09-10

    • The 2024-25 categories add to 119, not the 122 SFUSD stated. SFUSD's 50,028 students excludes charters and the 289 County Office students.
    • County and court schools jump from 6 to 9 in 2025-26 with no stated reason. CDE counts 4 County Office schools every year.
    • Counting rules differ: CDE's directory shows 102 non-charter K-12 schools plus 12 charters in FY25-26; SFUSD's website says 120 schools; its FY26-27 plan says 122 in one place and 101 in another. SFUSD counts 11 charters, Ed-Data 12. The State's 2025-26 files show 12 under SFUSD's code plus 2 under the State Board, so 14 in SF.
    • CDE shows 102 to 108 non-charter schools every year since FY09-10, while non-charter enrollment fell about 4,500 (est.). Average school size shrank.

    Sources: SFUSD Facts at a Glance, 2022-23 to 2025-26; Ed-Data non-charter schools, 2016-17 to 2025-26; Ed-Data charter enrollment.

    Four southeast ZIPs hold 42% of public school students and 54% of poor school-age children (est.)

    Show all ZIPs

    Proxy, not district data. These are Census survey estimates (ACS 2020-2024) of SF residents in any public K-12 school, charters included, by ZIP, with margins of about 9% to 16% in the eight largest ZIPs and up to about 60% in the smallest. SFUSD's 2023 Facilities Master Plan says the same four ZIPs are home to 47% of its students (fall 2021, PreK-12, no charters), but we found no official table by ZIP. In ZIPs with child poverty of 14% or more, about 79% of K-12 children attend public school, against 51% in ZIPs under 4% (est.).

    Share of SF's public K-12 students by ZIP; the label shows the school-age child poverty rate. Area names are approximate. Sources: ACS B14002 and B17001 via Census Reporter (raw data file); SFEDup, Sept 2024 (secondary) (district data by records request: the same four ZIPs hold over 45% of SFUSD students).

    Sources

    SFUSD, June 2026 (LCFF enacted in 2013-14); Ed-Data enrollment, 2009-10; Ed-Data enrollment, 2010-11 to 2015-16; Census school finance tables (FY2006 to FY2010 are separate files); Ed-Data chart 180, 2009-10 (3,347.5); Ed-Data chart 180, 2010-11 to 2015-16; Ed-Data chart 192, 2009-10 (16.5); Ed-Data chart 192, 2010-11 to 2015-16 (18.8 in 2010-11); Ed-Data County Office schools (4 every year, 2016-17 to 2025-26); Ed-Data non-charter schools, 2009-10 (104); Ed-Data non-charter schools, 2010-11 to 2015-16; Ed-Data charter schools (12 in 2025-26); Ed-Data charter enrollment, 2009-10 (2,193); SFUSD LCAP FY26-27 (122 schools in one place, 101 in another; 289 County Office students); Census Reporter table page, B14002; Census Reporter table page, B17001; 2023 Facilities Master Plan, Figure 41, PDF p. 57 (the same four ZIPs are home to 47% of SFUSD students, fall 2021).

    Sources

    Trend line: slope -0.84, r = -0.81, n = 99, 2026 tests (est.). The 2026 test file renames the measure "Percentage Proficient and Above." State test results 2019; State test results 2025; State test results 2026; State census day 2025-26; chronic absence 2018-19, 2024-25, 2025-26 (the State's SFUSD rows, with and without alternative-status schools); SFUSD 2023 Facilities Master Plan, PDF pp. 76-81 (its column is called Enrollment capacity); PDF p. 19 (Lowell and Ruth Asawa are admission-based high schools); Ed-Data teacher FTE (chart 180; this link shows Lowell, and each school has its own chart); State school directory; DataSF Analysis Neighborhoods; The Frisc, Feb 2026 (press) (strike); SFUSD data dashboard FAQ (State definition of socioeconomically disadvantaged); Education Code 60901 (chronic absentee: absent 10% or more of school days; public mirror).

    7 of 14 · Cuts

    SFUSD cut about $114M and about 375 jobs in a year and a half.

    About $114MAnnual spending cut over a year and a half, as of June 2026
    About 375Jobs cut in the same period
    15%Cut in central office spending in the same period
    3,387 to 2,947Teachers by SFUSD's own count, 2022-23 to 2025-26. Sources count teachers differently.

    Social workers, counselors, and aides were repeatedly proposed for cuts and saved with restricted or one-time money. That leaves them exposed as that money is spent down.

    Cut totals by budget year. These totals overlap, so do not add them.

    Show data table

    The gap returns in FY28-29: about $26M in unrestricted money, assuming voters renew the QTEA teacher parcel tax. Restricted funds run short in FY27-28 and FY28-29. "Restricted" money can only be spent on set purposes.

    See the data: Year by year, and the FY28-29 gap

    Social workers, counselors, and aides were repeatedly proposed for cuts and saved with restricted or one-time money, which leaves them exposed as that money is spent down. SFUSD projects restricted deficits in FY27-28 and FY28-29.

    FY24-25, about $103M in budget solutions

    Adopted June 25, 2024. Not all of it is service cuts.

    • AdoptedBooks and supplies budgets cut 55.3%.
    • AdoptedContracted services cut 23%.

    FY25-26, $113.8M in reductions

    Adopted June 24, 2025.

    • Adopted205 central office positions cut, $33.9M.
    • PreliminaryLayoff notices authorized in February 2025 for 46 counselors, 18 social workers, and 164 paraeducators.
    • RescindedFinal notices for 34 counselors and 117 paraeducators were rescinded in May 2025. We found no final count for social workers.
    • AvoidedLayoffs of permanent teachers, helped by 378 early retirements. Some temporary teachers were not renewed.

    FY26-27, about $102M cut target

    A fall 2025 target, replaced by the budget adopted June 23, 2026.

    • Not adoptedCutting 45 of 99 school social workers. Reversed December 18, 2025, using restricted money.
    • Not adoptedEnding busing for students outside special education. The bus contract was renewed.
    • AdoptedMostly vacancy savings, shifts to restricted funds, and lower contract spending.
    The gap returns in FY28-29

    SFUSD's June 2026 budget projects a gap of about $26M in unrestricted money in FY28-29. It projects restricted funds running about $66M short in FY27-28 and about $53M short in FY28-29, leaving about $117M. That projection assumes voters renew the QTEA teacher parcel tax, which otherwise ends June 30, 2028.

    These totals overlap, so do not add them. "Restricted" money can only be spent on set purposes. SFUSD's count covers teachers from transitional kindergarten through grade 12. Ed-Data's State staff series shows different totals that are not comparable.

    See the data: The cuts we found, adopted, proposed, or reversed

    Several of the harshest FY26-27 proposals were reversed; most of what passed was vacancy savings and fund shifts

    Rows overlap: totals, plans, and line items are listed together, so do not add amounts. Most FY26-27 "cuts" were vacancy savings, fund shifts, and budget cleanups. Social workers, counselors, and paraeducators were repeatedly proposed for cuts and saved with restricted or one-time money, which leaves them exposed as that money is spent down. A proposal to halve security staff has no separately reported outcome.

    Each row links its source. Main sources: SFUSD FSP Exhibit A, June 2024; Central Office Reduction Plan, April 2025; SFUSD 2026-27 Recommended Budget; SF Standard, June 2026 (press).

    Sources

    SF Standard, June 2026 (press; SFUSD's Facts page confirms $114M, and we found no SFUSD document for about 375 jobs or 15%); SFUSD Fiscal Stabilization Plan, June 2024; SFUSD, June 2025; Central Office Reduction Plan, April 2025; SFUSD Layoffs Overview, Feb 2025; SFUSD, May 2025; SFUSD Budget FAQs; Facts on SFUSD's Budget; Superintendent's message, Dec 2025; SFUSD 2026-27 Recommended Budget (PDF p. 5 budget changes; pp. 8, 131-133 projection; p. 160 QTEA); Mission Local, March 2026 (press; its $126M restricted figure was replaced by the June 2026 budget); SFUSD Facts at a Glance (teacher count); SPUR on QTEA (secondary); SFUSD FY25-26 budget deck, June 24, 2025 (PDF p. 20 positions on expiring restricted money; p. 22 staffing funded with one-time restricted funds; p. 27 Board dates); SFUSD, June 2026 (Board adopted the 2026-27 budget June 23, 4-3); SFUSD FY25-26 3rd Interim deck, May 12, 2026, PDF p. 7 (vacancy and budget savings); Ed-Data teacher FTE (chart 180), a different count.

    8 of 14 · Is it working?

    After inflation, spending grew to 1.2 to 4.6 times its starting level across all three systems (est.), and results are mixed.

    Reading is below 2019, the homeless count is above 2015, and overdose deaths are below their 2023 peak but above 2017.

    Schools

    Spending
    Up 106%

    Per student, after inflation (est.), FY05-06 to FY24-25.

    Results
    Reading 56.5% to 53.5%

    Share meeting the reading standard, 2019 to 2026.

    Homelessness

    Spending
    Up 360% (4.6 times)

    After inflation (est.), FY06-07 to FY24-25.

    Results
    6,775 to 7,973 people

    One-night count, 2015 to 2026. The 2026 count's method changed.

    Substance use treatment

    Spending
    Up about 20% (about 1.2 times)

    After inflation (est.), FY06-07 to FY24-25, on definitions that differ.

    Results
    625 deaths in 2025

    Accidental overdose deaths, 246 in 2017 and 810 in 2023.

    This shows a pattern, not a cause

    In our view, fentanyl, COVID, and housing costs also moved these numbers.

    See the data: Notes and numbers for each system
    Up 106%Schools, spending per student after inflation, FY05-06 to FY24-25 (est.)
    Up 360% (4.6 times)Homelessness, spending after inflation, FY06-07 to FY24-25 (est.)
    Up about 20%Substance use treatment, spending after inflation, FY06-07 to FY24-25, about 1.2 times its starting level on definitions that differ (est.)

    Schools

    Spending
    Up 106%

    Spending per student after inflation (est.), FY05-06 to FY24-25.

    Results
    Reading 56.5% to 53.5%

    Share meeting the reading standard, 2019 to 2026.

    • Math went from 50.6% (2019) to 45.8% (2026).
    • Average daily attendance fell 10%, from 50,233 to 45,086 (FY05-06 to FY24-25).
    Show the numbers

    Homelessness

    Spending
    Up 360% (4.6 times)

    Spending after inflation (est.), FY06-07 to FY24-25.

    Results
    6,775 to 7,973 people

    People counted in the one-night homeless count, 2015 to 2026. The 2026 count's method changed.

    • Unsheltered count of 3,400 in 2026, down 22% from 2024.
    • 93% of people who exited homelessness in FY23-24 did not return to the City's homelessness system within 12 months.
    • 771 of 9,160 supportive housing units were vacant in October 2026.
    Show the numbers

    Substance use treatment

    Spending
    Up about 20% (about 1.2 times)

    Spending after inflation (est.). FY06-07 to FY16-17 is actual spending on one DPH program (Substance Abuse - Community Care). FY20-21 on is DPH's budget for specialty services: $85M, $86M, $99M, $115M, and $125M after inflation, below the $104M of FY06-07 until FY23-24. Definitions differ, and DPH says some programs moved in from its mental health budget, so growth may be overstated.

    Results
    625 deaths in 2025

    Accidental overdose deaths, 246 in 2017 and 810 in 2023. 2017 to 2019 are DPH counts from death records; 2020 on are Medical Examiner counts.

    • People on methadone or buprenorphine die at about half the rate of when they are off it (meta-analysis, 2021).
    • 14% of people starting medication were still on it at six months (FY24-25).
    Show the numbers
    This shows a pattern, not a cause

    In our view, fentanyl, COVID, and housing costs also moved these numbers. Each chart uses one axis, and the three systems serve different people.

    See the data: Underused resources

    771 of 9,160 supportive housing units (8.4%) sat vacant in October 2026, above the 7% goal. Every reading we found since 2022 was above 7%; the closest was 7.1% in January 2024. Methadone programs ran well below the capacity DPH reported in FY21-22 to FY23-24, the last years it published capacity, and family shelters were 69% full on June 15, 2026. On our reading, the reasons for housing and methadone are mostly staffing, repairs, and rules rather than idle cash; we found no source explaining the empty family beds.

    Methadone slots were at most 57% to 68% filled

    Capacity is a point-in-time count. DPH called it contracted through FY22-23 and licensed in FY23-24 (2,504 served of 4,198, about 60%). People served is everyone treated in the year, so true fill was lower. Sources: DPH ToD FY21-22; DPH ToD FY22-23.

    Methadone patients rose 39% from 2022 to 3,141 in 2025, the highest in this series (2020 to 2025)

    People treated with methadone by calendar year. These calendar-year DPH billing counts do not match the capacity chart's fiscal-year people served (2,753 in FY21-22, 2,408 in FY22-23), which DPH takes from treatment admissions records. Even 2025's 3,141 would fill only about 75% of FY22-23 capacity (est.). Source: DataSF, DPH Substance Use Services (methadone rows).

    BAART's methadone contract has paid 45% of its ceiling with 93% of its term gone

    about 750Patients at BAART Market, 1111 Market St, in March 2023 (self-reported; 40% dosed daily)
    $98.3MContract ceiling, July 2018 to June 2027 (BAART Market and Turk)
    $43.8MPaid as of Oct 2026; $11.4M is encumbered and $43.1M remains

    City payments to BAART for methadone, by fiscal year ($M)

    DPH contract ceilings often include a 12% contingency, and on our reading they are set high for Medi-Cal-billed volume, so an unspent ceiling is not cash left idle. We found no current census or licensed capacity published for the clinic. Sources: DataSF Supplier Contracts; DataSF Vendor Payments; SF Standard, Mar 2023 (press); DHCS licensed programs.

    Of 771 vacant supportive housing units, 361 were offline and only 78 were open for referral

    Why 361 units were offline

    Site-based units, dashboard updated Oct 16, 2026; buildings in lease-up are excluded. The monthly rate ranged 8.1% to 9.1% over the past year, down from a 10.2% average in 2022-23. Sources: HSH PSH vacancy data; BLA audit, May 2024.

    Family shelters were 69% full while adult beds were 93% to 98% full

    Occupancy on June 15, 2026: 293 of 3,210 listed beds were empty. HSH's 4,438 budget figure also covers transitional housing, hotel vouchers, and other interim programs. Source: HSH Shelter Capacity and Occupancy Report.

    About 94 treatment beds were open on Oct 30-31, 2026; in FY24-25, only 51% of detox requests got same or next-day admission

      Sources: DPH bed finder (live counts; these are from Oct 30-31, 2026, no archive); SF Standard, Sept 2026 (press); DPH ToD FY24-25.

      Unspent money recurs: HSH spent 75% of its FY24-25 budget

      Sources: BLA 2024; BLA 2020; Controller spending data; Controller OCOH FY25; DPH Prop C update, Jan 2026; DataSF Supplier Contracts.

      Why capacity sits empty: on our reading, mostly operations, not idle cash

        The billing caveat

        Most contracted treatment is funded through Drug Medi-Cal or federal block grants, and Medi-Cal revenue follows units of service, so empty slots and unspent ceilings often mean money not drawn, not money paid for nothing. The fair claim is that funding is not turning into service at the rate paid for, not that the money is wasted.

        Sources: DPH ToD FY24-25; DPH ToD FY22-23; BLA 2024; SF Standard, Sept 2026 (press); SF Standard, Mar 2023 (press); DPH Prop C update; 42 CFR 8.12; Local News Matters, Aug 2024 (press, quoting DPH); SF Standard, July 2022, on turned-away patients (press); DPH update, Jan 2026.

        Sources

        BLA audit, May 2024, PDF p. 2 (7.1% on Jan 31, 2024); DPH Treatment on Demand FY23-24, Table 4, PDF p. 27; Board Resolution 234-22 (BAART contract ceiling and term); Health Commission contract report, Feb 2022, PDF p. 1 (12% contingency on a methadone contract under the same RFP as BAART); HSH budget overview, Jan 2026, slide 8 (4,438 beds by type); Controller budget data (HSH FY24-25, $834.6M); HSH PSH vacancy data; DPH bed finder (program eligibility); Business and Tax Code Sec. 2810 and 2811 (Prop C uses); DPH Treatment on Demand FY22-23, PDF p. 13 (most contracted services); DPH memo, Feb 2026, PDF p. 18 (revenue follows units of service).

        See the data: The drug crisis

        About 17,000 people in San Francisco need or use public substance use treatment, and the City's overdose death rate was about 2.7 times the national rate in 2025 (est.). Addiction and homelessness overlap heavily, but they are not the same population.

        About 17,000 people use or need public treatment; about 13,500 have opioid use disorder and about 15,000 to 17,000 use fentanyl regularly (est.)

        Best-supported range

        About 17,000 people with a substance use disorder serious enough to need or use the public treatment system: DPH treated 17,029 in FY24-25, and estimates 15,000 to 19,500 need it. For opioids and fentanyl, about 13,500 (opioid use disorder, survey) to 17,000 (regular fentanyl use, secondary model) (est.), range 7,000 to 26,000. Narrower counts measure smaller groups, such as the 7,839 people on medications for opioid use disorder.

        Highlighted rows are the best-supported figures. Sources: DPH Treatment on Demand FY24-25; Controller 2025 Needs Assessment; SAMHSA NSDUH 2021-2023; Facente et al. 2018; UCSF BHHI statewide survey, PDF p. 3 (35%); fentanyl user model (secondary, "planning numbers").

        San Francisco's overdose rate is about 3 times LA County's (est.)

        Show data table

        Overdose deaths per 100,000 residents, crude rates (est.) from CDC provisional county counts (12 months ending December) and Census population. Uses Census Vintage 2024; 2025 uses the 2024 estimate. Vintage 2025, used elsewhere on this site, moves these rates by 0.5 or less. CDC counts residents of the county (SF: 470 in 2025). The Medical Examiner (OCME) counts deaths that occur in SF, including non-residents and people without a verifiable address (625 in 2025), which gives about 75.5 per 100,000 (est.). Sources: CDC VSRR county counts; CDC VSRR national counts; Census estimates; OCME 2025.

        84% of homeless people with a behavioral health diagnosis have a substance use disorder

          Sources: Controller 2025 Needs Assessment; DPH ToD FY24-25; OCME 2025; 2024 PIT report.

          Most people who die of overdose are 45 or older

          Accidental overdose deaths by age, 2025. Six deaths of unknown age are not shown. Source: OCME 2025 report.

          Two in three people who died of overdose in 2025 had a San Francisco address, but an address does not mean City housing

          Fixed address in San Francisco
          415 (66%)
          No fixed address
          143 (23%)
          Fixed address outside San Francisco
          51 (8%)
          Unknown
          16 (3%)

          625 accidental overdose deaths, 2025. The Medical Examiner defines a fixed address as a verifiable address that "may indicate" residence, so it is not proof of where someone lived or who paid for it. Source: OCME 2025 report.

          How many were in City housing

          The Medical Examiner does not report this. DPH estimated about 26% of 2024 deaths occurred in supportive housing (cited by the Grand Jury). One advocacy count, method unverified, puts 959 of 4,090 deaths from 2020 to 2025 inside supportive housing, about 23%. Combining address and housing data suggests about 40% to 50% of those who died were homeless or in subsidized rooms. The rest includes private SRO rooms, other housing, addresses outside San Francisco (8%), and unknown (3%) (est., mixed years).

          Why some have housing

          Housing First has been City policy since 2004, and drug use does not disqualify anyone. Housing goes to those who score as most vulnerable. Before a July 2026 ordinance, 42 of about 8,500 to 9,200 supportive units were drug-free. SRO residents died of overdose at 19.3 times the rate of other residents (2010 to 2017), and 86% of them died at home.

          What the housing costs

          About $29,000 per unit a year ($421.6M for 14,498 units, FY25-26), about what SFUSD spends per student from all sources. 53% of single adults in HSH housing programs reported a substance use or mental health disorder. Applying that share gives very roughly $135M to $223M a year (est.). We found no source that separates drug use from mental health.

          Would people come or stay without services?

          We found no rigorous study that answers this. Under one year in SF rose from 6% (2019) to 32% (2024) among homeless people last housed in SF, and other California counties rose from 20% (2019) to 30% (2024). But 59% became homeless while living in SF, the out-of-state share is 11%, the same as 2015, and 3% to 4% of cash aid recipients had been in SF only 15 to 30 days.

          Records that would answer this

          Medical Examiner deaths by building, matched to the City's housing list. HSH data on how long people lived in SF before entering housing. Sheriff data on home addresses of RESET Center admissions.

          Sources: Rowe et al., 2019; The Voice of SF (advocacy); HSH budget overview, Jan 2026; Controller 2025 Needs Assessment; HSH 2024 PIT page; Civil Grand Jury, June 2026; OCME 2025; SF Examiner, Mar 2025 (press, reporting an HSA estimate).

          What the evidence shows on addiction, housing, and overdose deaths

          From peer-reviewed studies and official data. Each finding describes a pattern; it does not by itself prove a cause.

          Children and the drug market: about 3,000 to 3,500 children live in the Tenderloin, and Safe Passage helped more than 1,000 a day in 2023

            Funding for children

            The Mayor's proposed FY26-27 budget funds the Children's Baseline at $293.6M against a $260.4M requirement and grows DCYF by $23.0M. The Charter's deficit rule allowed the Student Success Fund pause, and the City chose it. In the same years the City added adult recovery beds and the RESET Center. DPH is ending its $973K school wellness line on June 30, 2027, after reviewing programs under the State's new behavioral health law, and no opioid settlement money has been found going to schools.

            Sources: Mission Local, Aug 2024 (press); AP, May 2024 (press); OCME 2025; OCME Oct 2026; Youth Commission Res. 2526-AL-12; DPH deck on the BHSA plan, Health Commission, Apr 2026; DPH opioid settlement memo, June 2024; Controller, June 2026.

            Sources

            Assaf et al., JAMA 2025 (37.1%); DPH Director's update, Sept 2025, slide 6 (2024 deaths by setting); 26% in supportive housing; BLA audit, May 2024, PDF p. 20 (2004 plan adopted Housing First); HSH Coordinated Entry Written Standards, Sept 2023, PDF p. 9; HSH Prioritization Memo, Aug 2026; Mission Local, July 2026 (press; 42 units; about 8,500 supportive units); HSH Point-in-Time survey dashboard, tab E, question E2 (2024: 32%, 414 respondents); 2019 count report, PDF p. 18 (6%); HSH FY26-28 ($832.7M); DPH, Substance Use Mortality Trends through 2024, Table 2; OCME 2023 report (810); OCME 2024 report (635); HSH, Jan 2026, slide 8 (54% in FY25-26); SFHSA Treatment Pathway flyer (Prop F, effective Jan 1, 2025); Mayor's release on the RESET Center, Jan 2026; National opioid settlement, Exhibit E, PDF p. 127; DPH budget submission, FY27 and FY28, PDF p. 11 (opioid settlement lines); Budget Book, PDF p. 127 (DCYF); Charter text, Sec. 16.131(d)(3), PDF p. 22; Controller, June 2025, PDF p. 34 (the FY25-26 pause); DPH Prop C budget deck, Aug 2026, PDF p. 2 (recovery beds).

            Sources

            After-inflation figures are FY25-26 San Francisco dollars, using SF-area consumer prices (est.). School spending is current expense of education per unit of average daily attendance. Homelessness is predecessor programs, then HSH, and leaves out county cash aid. State Current Expense of Education; Bureau of Labor Statistics, SF-area consumer prices; State test result files; Controller spending data; DPH Treatment on Demand FY24-25; Budget and Legislative Analyst audit, 2020 (earlier counts); HSH 2026 count; Home by the Bay, Year 2; HSH vacancy data; Medical Examiner 2025 report; Santo et al., 2021; DPH, Substance Use Mortality Trends through 2024, Table 2, PDF p. 41; Medical Examiner annual overdose reports, 2020 to 2025; State test results for SFUSD, 2019; State test results for SFUSD, 2025; State test results for SFUSD, 2026; 2024 count report, PDF p. 17 (2022 and 2024 counts); HSH deck on the 2026 count, slide 8 (method changes); DPH Treatment on Demand FY21-22, Table 1 (FY20-21 budget); DPH Treatment on Demand FY22-23, Table 4 (FY21-22 and FY22-23 budgets).

            9 of 14 · What money can do

            At this price and scale, tutoring could move a student at the middle of the class about 2 to 6 percentile points, and a student further behind about 1 to 5 (est.). $50M spread evenly moves the average student about 1 (est.).

            Spread evenly
            $50M a year

            About $1,000 per student. Sustained four years, about 1 percentile point (est.).

            High-dosage tutoring
            $9.9M a year

            About 4,000 to 8,000 students at $1,200 to $2,500 each, starting with about 3,600 at the highest-poverty schools (est.). About 2 to 6 percentile points for a student at the median (est.).

            Attendance outreach
            About $0.1M a year

            Could add about 0.5 to 1.1 days per student. State money follows slowly: none in year 1 (est.).

            Targeted money buys more test-score gain per dollar than money spread evenly (est.). So the ask names the programs, who comes first, and the results to report.

            Test score gain for a student in the middle of the class, in percentile points (est.)

            Low to high rangeMidpointSingle estimate

            Limits. These effects come from other cities and are not a promise, and tutoring effects shrink at large scale. General spending also shows lasting gains that this comparison leaves out.

            See the data: Costs, terms, and attendance money

            Targeted money buys more test-score gain per dollar than money spread evenly (est.), so the ask names the programs, who comes first, and the results to report.

            Spread evenly
            $50M a year

            About $1,000 per student today, about $760 in the study's 2018 dollars. Sustained four years, that adds about 0.02 standard deviations, roughly 1 percentile point (est.).

            High-dosage tutoring
            $9.9M a year

            Tutoring for about 4,000 to 8,000 students at $1,200 to $2,500 each (est.), starting with about 3,600 at schools over 75% low-income. That's about a third of all students at those schools, and about the number below the reading standard there (3,634 in tested grades, est.). The largest recent programs at this price gained about 0.04 to 0.09 standard deviations, with single sites up to 0.16, and one found no gain in math. We use 0.05 to 0.15, about 2 to 6 percentile points for a student at the median and less for a student further behind (est.). The 0.16 to 0.22 found in randomized trials is what a strong program could reach.

            Attendance outreach
            About $0.1M a year

            Could add about 0.5 to 1.1 days per student. State money follows slowly: none in year 1, and $0.6M to $1.26M a year only from year 4 (est.).

            How attendance money follows. SFUSD is funded on a three-year attendance average that runs 451 to 890 students above projected actual attendance, so year 1 brings no new money, year 2 about $0.2M to $0.4M, year 3 about $0.4M to $0.8M, and $0.6M to $1.26M a year only from year 4, if the mailers continue and keep working (est.). That counts every chronically absent district-run student. At the schools over 75% low-income where Part 3 starts, it is about 4,600 students, about $31,000 of mailers, and at most about $0.4M a year (est.).

            San Francisco paid for a lighter version in 2025-26. More than 2,700 students at 20 SFUSD schools got short reading-tutoring sessions in class, after DCYF used Student Success Fund money in January 2026 to add 1,443 students, with a focus on grades 2 to 4. The SF Education Fund runs it with Chapter One, a tutoring nonprofit, for about $500 a student a year (press reports). Its 2026-27 funding is not confirmed. A four-year randomized trial of the same program, in another district, offered kindergartners up to four years of support: students offered it scored 0.12 to 0.17 standard deviations higher on State reading tests in grades 1 to 3 (working paper, 2026).

            What a standard deviation is. It is how researchers compare test gains across studies. A gain of 0.10 standard deviations moves a student from the 50th percentile to about the 54th, and from the 20th to about the 23rd.

            Why doubled spending hasn't shown up in scores. Part of the rise came from fewer students being counted, and much of it went to benefits, pensions, special education, and contracts, which test scores do not measure (est.).

            Limits. These effects come from other cities and are not a promise, and tutoring effects shrink at large scale. General spending also shows lasting gains in graduation, college, and earnings that this comparison leaves out.

            See the data: What more money would do

            In this section, SD means a standard deviation of test scores, which is how researchers compare test gains across studies. A gain of 0.10 SD moves a student from the 50th percentile to about the 54th.

            One SD of test-score gain costs about $8,000 to $50,000 per student through tutoring, against about $165,000 through general spending in today's dollars (est.)

            Targeted programGeneral spendingMidpoint, or a single estimate

            Cost per student divided by effect size (est.). Not apples to apples: general spending also shows lasting gains in graduation, college and earnings, while tutoring's long-run effects are less studied and may fade. Summer cost is in 2014 dollars. Sources: Personalized Learning Initiative, June 2025; Nashville, Aug 2024; Kraft working paper, Oct 2024; Jackson & Mackevicius working paper (2018 dollars); Jackson & Mackevicius 2024; NEPC review of Reber & Goodman 2025; Kraft, Schueler & Falken 2026; Guryan et al. 2023; RAND summer learning 2016.

            Spread evenly, $50M a year moves average scores about 0.02 SD (est.)

            $50M

            All results are estimates (est.) with ranges. Effect sizes come from other cities, so they are not a promise. Only students in tested grades (3 to 8 and 11) show up in State tests. Tutoring effects shrink at scale: 0.29 SD pooled across trials, 0.16 to 0.22 SD in large randomized trials, and 0.04 to 0.09 SD in the largest recent evaluations. This uses 0.05 to 0.15 SD. Attendance revenue assumes D.C.'s preliminary result carries over: students were 6.9% less likely to be absent on days with a session. At three sessions a week that is about 1.4 more days a year, and about 3.1 for students who had missed 30% or more of school (est.). Philadelphia's mailer effect is added. Spending effects assume the money is sustained four years and scale in proportion. The skeptical reading (Reber & Goodman, Brookings 2025) finds about one-third of the meta-analytic effect using state-level correlations; NEPC disputes that method.

            Inputs: about 50,000 district-run students; 2,996 homeless students at district-run schools (FY25-26; 3,624 counting charters), 19.1% meeting the math standard and 25.8% reading (FY24-25); about 13,700 chronically absent district-run students (27.8% of 49,224, 2025-26); about $83.50 of State formula money per student-day (est., $677.5M over 45,086 attendance, 180 days). Sources: State chronic absence 2025-26; SFUSD 2026-27 budget; Jackson & Mackevicius 2024; Kraft et al. 2026; Rogers & Feller 2018; National Student Support Accelerator on D.C. tutoring; Chalkbeat on D.C. tutoring (press; its two and five days assume a session every school day); Ed-Data homeless students; Ed-Data reading results, homeless students (25.81%, 2024-25); Ed-Data LCFF; NEPC review.

            16 of 36 interventions rest on strong causal evidence

            Strong = randomized trial, lottery, or meta-analysis of causal studies; Moderate = single quasi-experimental study; Weak = correlational. SD = standard deviation of test scores; pp = percentage points. Costs marked as our estimates use SFUSD unit prices. Each study is linked in its row. For the mailed absence reports: Rogers & Feller supplement; vendor slides, not peer reviewed; Stanford business school, April 2024 (secondary); SFUSD's 2026-27 plan. For the emergency cash study: the authors' lab.

            Why SFUSD's doubled spending hasn't shown up in scores

            An illustrative gap (est., not a causal test)

            Real spending per student rose about $7,200 from FY18-19 to FY24-25 (est.). At the research average of 0.0316 SD per $1,000 in 2018 dollars, about $5,650 in those dollars predicts about +0.18 SD. SFUSD's proficiency fell slightly instead. Caveats: COVID and costs such as pensions, retiree health, and premiums that never reach classrooms.

              Why the ask is built this way

              Money that funds benefits growth, mandated costs, and contracts is less likely to show up in test scores. The strongest test-score evidence here is for tutoring, with more conditional evidence for smaller early classes and effective teachers. So the ask funds targeted, tracked dollars: a named program, named students, a dosage standard, and an outcome report.

              Sources: CDE Current Expense of Education; Ed-Data ELA; Ed-Data math; Ed-Data chronic absence; Ed-Data high-need share; Ed-Data spending by object; Ed-Data spending by function; SFUSD Unaudited Actuals 2023-24 (PDF p. 131 special education; p. 17 contributions); BOS Res. 495-24; BLA 2023; KQED, June 2025 (press); SFEDup (secondary) (secondary analysis); SFUSD, Oct 2025; SFUSD PEEF evaluation, 2022-23; DCYF evaluation RFP, April 2026; Hough & Loeb 2013; The Frisc, July 2024 (press).

              Sources

              Bhatt et al., NBER w32510, 2024, Section 6.3 (some fade-out six months after tutoring); SFUSD Facts at a Glance, 2025-26 (50,028 students; 15,048 in grades 9-12); Education Code 60640 (tests in grades 3 to 8 and 11; public mirror); State census day 2025-26 (2,996 homeless students at non-charter schools); Ed-Data math results, homeless students (19.07%, 2024-25); State test results 2019; State test results 2025; State test results 2026; State chronic absence 2018-19 (14.2% at district-run schools); State chronic absence 2024-25 (23.9% at district-run schools); SFUSD 2026-27 budget, PDF p. 161 (contributions go to special education and routine maintenance); Ed-Data high-need share, 2013-14 (67.01%); DCYF, Jan 20, 2026 (tutoring expansion, Student Success Fund); The 74, Feb 2026 (press; about $500 a student); Jung et al., four-year trial of Chapter One, EdWorkingPapers 26-1548 (working paper); Evidence for ESSA, Chapter One.

              Sources

              Percentile points assume normal test scores and a student at the 50th percentile (est.). Tutoring costs $1,200 to $2,500 per student a year. Outreach is mailers at $6.60 (2014-15 dollars, about $9 today) for the households of about 13,700 students, 0.5 to 1.1 days each, at about $83.50 of State formula money per student-day. SFUSD's three-year funding average spreads that money: $0 in year 1, one-third in year 2, two-thirds in year 3 (est.). Personalized Learning Initiative, June 2025; Nashville, Aug 2024; Education Code 42238.05 (three-year average; public mirror); State chronic absence 2025-26; SFUSD 2026-27 budget (PDF p. 5, funded and actual attendance); Jackson & Mackevicius 2024; Kraft, Schueler & Falken 2026; Nickow, Oreopoulos & Quan 2023; EdSurge, Oct 2026 (press; it credits Annenberg, and we did not find the range in the paper it links); Rogers & Feller 2018; Ed-Data LCFF; State Current Expense of Education; Ed-Data spending by object; BLS CPI-U, national; BLS CPI-U, San Francisco area.

              10 of 14 · The ask

              $15.5M a year in new money, plus protection for school money the City already promised.

              New money

              1. 1
                School-based drug prevention and school mental health staff.
                $3.0M a year
                Paid fromOpioid settlement funds
                Targeted toSchools near the Tenderloin and SoMa drug market first

                SF's own settlement terms need a legal read, so the City Attorney should confirm. This money is shrinking, so DPH has to plan for it. In fall 2024, 8.9% of SFUSD high schoolers were offered, sold, or given an illegal drug at school (student data).

              2. 2
                Youth early intervention at school sites.
                $2.5M a year
                Paid fromBehavioral Health Services Act (Prop 1)
                Targeted toStudents showing early signs of substance use or mental health needs, or at high risk from trauma or homelessness

                County money can't pay for broader prevention, which now runs through a State program. In fall 2024, 1 in 4 SFUSD high schoolers felt sad or hopeless for two weeks or more (student data).

              3. 3
                Tutoring and attendance for students below standard, starting at the highest-poverty schools.
                $10.0M a year
                Paid fromGeneral Fund freed by moving $10.0M a year of HSH supportive housing costs to Prop C
                Targeted toStudents below the reading or math standard, starting at schools over 75% low-income, where scores are lowest

                No current program would be cut, but Prop C would have $10.0M a year less for new housing, renewals, and expected federal cuts. The Controller and City Attorney need to confirm it meets Prop C's rules. A payment to SFUSD needs a data-sharing agreement. If the swap fails that review, the fallback is $10.0M of the Student Success Fund's scheduled FY27-28 increase. That is not new money, and it could reach only schools that meet the Fund's rules.

              Protection, not new money

              1. 4
                Keep the Student Success Fund on schedule in FY27-28 and FY28-29.
                Protects up to about $25.0M, then $26.8M what a cut to the $35M floor would cost in FY27-28, then FY28-29 (est.)
                Paid fromGeneral Fund, under the Charter schedule
                Targeted toNo freeze or cut in FY27-28, or in FY28-29, when SFUSD's gap returns

                Each March's deficit test (about $220M last cycle) decides if the City may freeze or cut. A simple freeze would cost less.

              2. 5
                If the City freezes or cuts the fund, use its reserve to keep money flowing to students.
                Up to $18.1M, one time only as far as that year's Charter amount allows
                Paid fromStudent Success Fund reserve
                Targeted toStudents, for the fund's purposes: school wellness and academics

                Only in a year the March deficit test is met, or extra local property tax (Excess ERAF) falls by half. City Attorney should confirm.

              The $15.5M a year sits in the lower part of the $8M to $43M a year we estimate its three channels could carry (est.).

              Where the $15.5M a year comes from

              What SFUSD would do in return

              Match with its own funds at the same schools, not cut its spending there, and report results every quarter. This match is being requested and is not yet secured. It is a three-year pilot.

              Limits on what can move. Within this pool, going higher means voters changing Our City, Our Home (Prop C) or deep cuts to housing and treatment. City deficits claim any savings first.

              See the data: Full notes on each part

              New money

              1. 1
                School-based drug prevention and school mental health staff.
                $3.0M a year
                Paid fromOpioid settlement funds
                Targeted toSchools near the Tenderloin and SoMa drug market first

                National settlements list school-based prevention, and mental health services by school staff that lower the risk of drug misuse, as approved uses. SF's own settlements, most of the dollars, have separate use clauses we have not seen, so the City Attorney should confirm. DPH says this money was front-loaded, projects about $4M to $5M of fund balance remaining after FY27-28, and expects further declines after FY29-30, so a three-year line needs DPH to plan for it. Programs that combine social skills and resistance training had small but consistent protective effects against drug use in a 2014 review of 51 trials; the one trial of an information-only program found no effect (Cochrane review, 2014). In fall 2024, 8.9% of SFUSD high schoolers were offered, sold, or given an illegal drug at school (student data).

              2. 2
                Youth early intervention at school sites, with the $973K of school wellness programs DPH plans to end June 30, 2027 re-scoped to serve students showing early signs, because county Behavioral Health Services Act money can't pay for broader prevention.
                $2.5M a year
                Paid fromBehavioral Health Services Act (Prop 1)
                Targeted toStudents showing early signs of substance use or mental health needs, or at high risk from trauma or homelessness

                Money for broader prevention, including school-based prevention, now runs through a State Department of Public Health program. In 43 trials, mental health services that school staff gave elementary students reduced problems, most for students already showing problems (Sanchez et al., 2018). In fall 2024, 1 in 4 SFUSD high schoolers felt sad or hopeless for two weeks or more (student data).

              3. 3
                Tutoring and attendance for students below standard, starting at the highest-poverty schools.
                $10.0M a year
                Paid fromGeneral Fund freed by moving $10.0M a year of HSH supportive housing costs to Prop C
                Targeted toStudents below the reading or math standard, starting at schools over 75% low-income, where scores are lowest

                This is a funding swap. Prop C (the Our City, Our Home Fund) would pay $10.0M a year of supportive housing costs the General Fund pays now, and the freed General Fund would pay for Part 3. That's about 1.3% of HSH's budget and about 2.4% of its Prop C plan in FY27-28, the first year this could start. HSH has proposed a smaller version: in its February 2026 budget request, it met a $4M General Fund target partly by moving a family shelter to Prop C. No current program would be cut. But Prop C would have $10.0M a year less for new housing, renewals of time-limited subsidies, and expected federal cuts. The Controller and City Attorney need to confirm it meets Prop C's rules, including its limits on replacing City funding. A payment to SFUSD also needs the data-sharing agreement in Charter Sec. 16.133. Fallback. If the swap fails that review, Part 3 could come from the Student Success Fund's own increase. The adopted budget raises the Fund from $45.0M in FY26-27 to $60.0M in FY27-28. The Charter's purposes for the Fund include grade-level success in core academic subjects, and it names in-classroom tutors (Charter Sec. 16.131). On our reading, the route is a District Innovation Grant from DCYF to SFUSD for a program at a group of schools, in an amount DCYF or the Board of Supervisors sets. Limits: on the same reading, every school covered needs a School Site Council that endorses the plan and a full-time Community School Coordinator, in place or being hired, so the fallback could reach only schools that meet those rules. $10.0M would be larger than the Fund's whole District Innovation and Workforce Pipeline line in FY24-25 ($7.25M, of which $2.8M was spent; Fund annual report, PDF p. 32). The Fund's FY25-26 budget put $11.05M in District Innovation (same report, PDF p. 33). The Fund has paid for tutoring: in January 2026, DCYF used Student Success Fund money to add 1,443 students to reading tutoring at SFUSD schools (DCYF, Jan 20, 2026). It is not new money: the City has already scheduled it for the Fund, so the ask's new money would fall from $15.5M to $5.5M a year. Prop C would be untouched. It exists only if the City does not freeze the Fund (Part 4), and that test comes in March 2027, after departments submit budgets in late February. How much of the increase is already committed is a question for DCYF.

              Protection, not new money

              1. 4
                Keep the Student Success Fund on schedule in FY27-28 and FY28-29.
                Protects up to about $25.0M, then $26.8M what a cut to the $35M floor would cost in FY27-28, then FY28-29 (est.)
                Paid fromGeneral Fund, under the Charter schedule
                Targeted toNo freeze or cut in FY27-28, or in FY28-29, when SFUSD's gap returns

                The March 2027 update tests FY27-28, and the March 2028 update tests FY28-29, each against a deficit threshold (about $220M last cycle). A freeze in FY27-28 would hold the fund at $45.0M, $15.0M below the $60.0M budgeted, and could lower where FY28-29 starts, a question for the City Attorney. A simple freeze in FY28-29, at $60.0M, would cost about $1.8M to $2.9M. The only freeze so far, FY25-26, held the fund $10.0M below schedule. The $26.8M uses the Charter's 3% growth cap; the Controller's March 2026 table implies about $27.9M, and the Controller should confirm.

              2. 5
                If the City freezes or cuts the fund, use its reserve to keep money flowing to students.
                Up to $18.1M, one time only as far as that year's Charter amount allows
                Paid fromStudent Success Fund reserve
                Targeted toStudents, for the fund's purposes: school wellness and academics

                The Charter lets the City draw this reserve only in a year the March report shows a deficit above the threshold (about $220M), or the Controller finds Excess ERAF (extra local property tax returned to the City) will fall by half or more. In the Excess ERAF case, the reserve must be used first toward the $35M floor. Total spending from the fund can't pass the Charter amount for that year. On the Controller's schedule, a freeze leaves room for about $15.0M in FY27-28 and about $2M to $3M in FY28-29; a cut toward $35M leaves more. Where unspent draws go is a question for the City Attorney, who should confirm all of this. The $18.1M is the last reported balance, from June 30, 2025; the June 30, 2026 balance is not published and could be higher.

              The $15.5M a year sits in the lower part of the $8M to $43M a year we estimate its three channels could carry (est.). Across all channels the estimate is $13M to $87M, but $5M to $22M of that pays only for homelessness services for families and youth, not school programs. Part 3 frees $10.0M of General Fund by moving HSH supportive housing costs to Prop C. That's about 1.3% of HSH's budget and about 2.4% of its Prop C plan in FY27-28, the first year this could start.

              Where the $15.5M a year comes from

              Part 5 depends on a Charter rule that ties reserve draws to years the March deficit test is met or Excess ERAF falls by half; the City Attorney should confirm. Parts 4 and 5 protect money and add no new money. SSF is the Student Success Fund. HSH is the Department of Homelessness and Supportive Housing. DPH is the Department of Public Health.

              What SFUSD would do in return
              • Match with its own funds at the same schools.
              • Not cut its spending at those schools.
              • Report quarterly on chronic absence at funded schools, students tutored, local interim test scores, dollars spent by school, and any reserve drawdown. Report State test scores once a year.

              It is a three-year pilot. Each year's money still needs a new appropriation. This match is being requested and is not yet secured.

              Who controls the money

              Low to high rangeMidpoint, or a single estimateCity poolCould reach students

              Within this pool, going higher means voters changing Our City, Our Home (Prop C) or deep cuts to housing and treatment. Money from outside this pool, such as new revenue or other City departments, is a separate question.

              The first three steps narrow General Fund money, though the realistic step also counts $2M to $8M of opioid money from outside the pool. Restricted channels are added on top, but the rows are not independent. HSH's FY26-28 request met a $4M General Fund target by moving shelter costs to opioid and Prop C money. Part 3 uses the same kind of swap, so it competes with those rows and with the City's deficit. Years are mixed (FY24-25 to FY26-27), so read these as orders of magnitude. The gross pool leaves out about $29.8M of opioid settlement money usable only within its rules, about $472M with it. City deficits claim any savings first: FY27-28 balances with about $458.3M of one-time fund balance and about $73.5M of reserves (adopted), and the Controller's June 2026 review put FY28-29 at about $450M (preliminary). Legal readings are from public documents, and the City Attorney should confirm each path.

              See the data: Build the ask

              The default is the ask above: a three-year pilot, targeted to the highest-need students, with a requested SFUSD match. The total counts only new money per year. Parts 4 and 5 protect money and are shown separately. Adjust each part to see the total and what it buys.

              Unit costs are FY26-27 estimates, except the FY25-26 ELO-P rate: salaries from the 2025-27 UESF certificated and classified schedules plus SFUSD's 48.8% benefit ratio ($359.0M of benefits ÷ $735.2M of salaries); tutoring from recent evaluations: about $1,200 to $2,000 in the University of Chicago's Personalized Learning Initiative (p. 8, 2023 dollars) and $1,500 in Nashville's district-run program (PDF p. 39), within the $1,000 to more than $3,000 Accelerate reports across providers (PDF p. 16); the $1,200 to $2,500 range used for the count is from EdSurge, Oct 2026 (press, citing Annenberg; we did not find the range in the paper it links); after-school seats at the State ELO-P rate.

              Sources

              SFUSD 2026-27 Recommended Budget, Form MYP, PDF p. 134 (benefits $359.0M ÷ salaries $735.2M = 48.8%); Ed-Data, average teacher salary (chart 199): $103,472 in FY24-25; UESF classified salary schedules (Schedule 935, paraeducators); EdSurge, high-impact tutoring costs (Oct 2026) (press, citing Annenberg; we did not find the range in the paper it links).

              See the data: Who controls the money

              General Fund and opioid settlement rows move by simple majority, and most carry service or matching-fund losses

              Amounts are the latest published year for each row; "(est.)" marks calculations. Sources are linked in each row. Opioid settlement use rules: SF Standard, Sept 2023 (press) (Controller on each settlement's own eligible uses); Settlement Exhibit E.

              Legal conclusions need the City Attorney

              These are readings of public documents, not legal advice. Open questions: whether the reserve can be drawn in a year the deficit test is met but the City doesn't freeze or cut (the Charter's words are "any fiscal year described in" the freeze and Excess ERAF subsections, so it can't be drawn in other years); whether Care Not Cash (Prop N, 2002) can change without voters; the vote needed for a Board-placed Prop C amendment; legal floors for Jail Health and county cash aid; gift-of-public-funds findings for City grants to SFUSD; whether SFUSD has the data-sharing agreement Charter Sec. 16.133 requires for payments the Charter or State law doesn't require; and each opioid settlement's use clauses.

              The simplest route, on our reading

              An add-back to DCYF for an SFUSD grant, with public-purpose findings, conditions, and the data-sharing agreement Charter Sec. 16.133 requires. In late June the Board can move money only within the Mayor's fund totals, with final votes in July. The Mayor can then reduce or reject it unless two-thirds of the Board restores it, and we found nothing that makes the Mayor spend it. So the June 1 proposal matters most. See Calendar.

              About the SF data sites · secondary, independent sites

              Two of 11 independent SF data sites we reviewed carry City dollar figures; of 54 key figures, 15 are confirmed, 8 consistent, and 31 unverified

              An independent directory lists 20 SF civic data tools (built on manus.space). We reviewed the 11 below, read each site's embedded data, and checked its key figures against the primary sources behind this site. None is an official City product. "SF Drug Strategy" is now an advocacy deck for arresting public drug use and expanding the RESET Center, so treat its framing as one side.

              Show every figure checked
              ✓ Confirmed matches a primary source≈ Consistent agrees within rounding or method? Unverified no primary checked
              How we use them

              We cite dollar figures to the City primaries (Controller, HSH, DPH), not to these sites. Several sites cite each other, so agreement among them is not independent confirmation. The "17% from SF" street-interview figure describes people interviewed on the street. The 2024 count (59% became homeless while living in SF) and overdose records (8% of 2025 decedents had an address outside SF) measure different things, so they don't test it.

              Secondary sites, as published through Oct 30, 2026: . Checked against HSH budget overview, DPH Treatment on Demand FY24-25, Controller OCOH FY25, DPH deck on the BHSA plan, Health Commission, Apr 2026, and Controller spending data.

              Sources

              Board budget hearing calendar, 2026, p. 4 (committee sets add-backs June 25; votes July 21 and 28); SF Data Sites directory (secondary; lists 20 tools); HSH 2024 Point-in-Time Count page (59% became homeless while living in SF); OCME 2025 Overdose Report (51 of 625 had a fixed address outside SF).

              See the data: Where the money is

              Two of the 11 funds listed here have rules that let money reach students, on our reading; six more can in part, and PEEF already does; most homelessness and behavioral health money is locked by voters, the State, or federal grantors. The streams whose rules name youth or schools are where this ask starts.

              Two of 11 funds can reach students, on our reading

              Legal readings are from public documents, not legal advice. The City Attorney should confirm each path.

              Sources

              $10.0M of $780.3M is 1.3% (FY27-28 adopted). Of the $421.68M HSH's Prop C plan spends in FY27-28, it is 2.4%, and that plan's sources equal its uses that year (HSH, June 2026, PDF pp. 10 and 18). The 2026 shelter swap is on PDF p. 2 of HSH's budget memo, February 2026. The Charter caps growth under its formula at 3%. In FY27-28, $60.0M less the $35M floor is $25.0M, and less a $45.0M freeze is $15.0M. Starting from the $60.0M the Controller schedules for FY27-28, FY28-29 is no more than $61.8M and Part 4 about $26.8M. Read literally, the same cap applied to FY26-27's $45.0M appropriation would hold FY27-28 to about $46.35M, so the Controller and City Attorney should confirm the base. Controller Revenue Letter, June 2026 (FY28-29 deficit, p. 5; reserve, pp. 27 and 31; schedule and the $220.0M freeze threshold, p. 38; Prop C revenue, $374.6M eligible spending, and $200.8M base, pp. 38-39); Mayor signs the budget, July 2026; Controller March 2026 Update, Appendix E (the $62.9M for FY28-29, and earlier deficit projections); Ordinance 140-26, the final budget: fund balance and reserves, p. 13; HSH, pp. 17 and 130; Student Success Fund, p. 67; Care Not Cash, pp. 130-131; Jail Health and DPH behavioral health, p. 178; opioid and BHSA, p. 181; Charter Sec. 16.131; Sec. 16.133; Sec. 9.103; Sec. 9.104; Sec. 9.118; California allocation agreement, Sec. 4.B.ii(b), p. 4 (half of Abatement Accounts money to listed uses, including vulnerable youth); Welfare and Institutions Code 5892 (choose the version amended by Stats. 2026, Ch. 27): (a)(3)(B) early intervention shares; (c) transfer caps; (f)(1)(E) State Department of Public Health prevention, including school-based programs; HSH budget submission, pp. 2 and 5 (met the $4M target by moving shelter costs to opioid and Prop C money); DPH budget submission, p. 2 (opioid money front-loaded; about $4M to $5M of fund balance left after FY27-28) and p. 11 (opioid lines; Jail Health revenue); Nine-Month Report, FY25-26, p. 29 ($18.1M at June 30, 2025) and p. 7 ($200.8M Prop C base); Charter text (File 240545): PEEF approval and reserve rules, pp. 5-6 and 11; Student Success Fund, pp. 21-23; National opioid settlement agreement, Exhibit E, Schedule B, items G.8 and G.12, p. E-12 (PDF p. 127); DPH deck on the BHSA plan, Health Commission, Apr 2026: slide 16 ($92.0M), slide 18 ($973K, four programs, June 30, 2027), slide 27 (primary prevention programs ended); HSH Budget FY26-28; Mayor's Proposed Budget Book; City Attorney, May 2023 (settlement dollars by defendant); SF Standard, Sept 2023 (press, quoting the Controller: each settlement has its own eligible uses); DHCS BHSA County Policy Manual, Ch. 7, Sec. A.7.1 (indicated prevention is the only prevention allowed under early intervention); Welfare and Institutions Code 5840 (choose the version amended by Stats. 2024, Ch. 40): who early intervention can serve; SFUSD 2026-27 Recommended Budget, Form MYP, PDF pp. 8 and 131 (FY28-29 gap of about $26M in unrestricted money); Controller, FY25-26 and FY26-27 Revenue Letter (June 2025), p. 34 (growth paused; $35.0M in FY25-26, against $45M in the Charter); Charter Sec. 9.113(f) (biennial appropriations expire at the end of the budget cycle); Measure E legal text (passed Nov 3, 2026, preliminary); SF Elections, Nov 3, 2026 results.

              11 of 14 · Questions

              17 questions, answered with numbers and sources.

              Open any question. Where a hard objection is true, the answer says so.

              Common questions

              Who pays for SF's public schools?

              The State sets most of it by formula, while State cash is about 46% and local money about 50% (est.). In FY26-27, State formulas and grants set about 70% of the $1.38B in revenue for SFUSD and its County Office, other local sources are 26%, and federal is about 4%.

              More

              About $329M of the formula is local property tax, which is why State cash is lower (SFUSD general fund). SFUSD 2026-27 budget, Form 01; State Budget Overview

              Is SFUSD part of the City?

              No. SFUSD is a separate government with its own elected Board of Education. The Charter requires about $167.9M for it in FY26-27 (est.). The adopted budget appropriates about $177.8M on those lines, because it also spends unspent PEEF balances.

              More

              DCYF's request budgets about $5.5M of the Student Success money for overhead and contracts, not grants. City grant agreements with SFUSD have been capped below the appropriation: not to exceed $29.0M of $35.0M in FY25-26, and $26.5M of $35.0M in FY24-25. In FY24-25 SFUSD spent $21.7M of $34.8M allocated, after the City added about $8.5M during the year (Student Success Fund annual report, PDF p. 32). Ordinance 140-26, PDF pp. 67-68; DCYF budget submission; Res. 471-25; Res. 483-24; Controller, June 2025; Charter Sec. 8.100 (elected Board of Education); Controller, June 2026, PDF p. 38; Budget Book, PDF p. 125 (unspent PEEF balances); Controller, June 2024, PDF p. 31 ($35.0M in FY24-25)

              Can the Mayor and Board just move this money?

              Only part of it. The Mayor and a Board majority control about $442M in this area (est., mixed years), and a realistic redirect is about $7M to $58M a year (est.). Our City, Our Home money must stay on homelessness unless voters change it. Who controls the money Business and Tax Code Sec. 2810 and 2811 (Prop C uses)

              How much does SF spend on homelessness and behavioral health?

              About $1.53B in FY26-27 budgets (est.), which is $832.7M for the homelessness department (HSH) plus $698.2M for Public Health's behavioral health division, both adopted. Most of the behavioral health money is mental health. Actual spending runs lower, and HSH spent $629.8M of $834.6M in FY24-25.

              More

              HSH budget page, FY26-28; Ordinance 140-26, the final budget; DPH Treatment on Demand FY24-25, Table 2 ($120.9M is substance use); Controller spending data (HSH FY24-25, $629.8M); Controller budget data (HSH FY24-25, $834.6M)

              How much goes to addiction treatment?

              Specialty treatment was $120.9M in FY24-25. Adding other substance use lines gives about $123M, and adding partly related crisis, jail health, and prosecution costs gives up to about $215M a year (est.).

              More

              DPH Treatment on Demand; SFFD Budget Book FY25 and FY26 (overdose, crisis, and EMS-6 lines); DPH Jail Health deck, 2025, PDF p. 8 ($55.4M); BLA memo, Apr 2019 (22.5% drug-case share, FY17-18)

              How many people in SF have a substance use disorder?

              DPH treated about 17,000 people in FY24-25 (17,029) and estimates 15,000 to 19,500 need public treatment. About 13,500 have opioid use disorder and about 15,000 to 17,000 use fentanyl regularly (est., secondary model). A national survey that includes alcohol and cannabis counts about 160,000 residents 12 and older (est.).

              More

              DPH Treatment on Demand FY24-25; federal drug use survey; fentanyl user model (secondary, planning numbers); The drug crisis

              How do cash aid and housing rules work for someone new to SF?

              County cash aid requires 15 verified days in SF, or 30 days for its job program, PAES. There is no housing entitlement, and 10% to 20% of assessed adults qualify for supportive housing.

              More

              In the 2024 count, 11% of homeless residents were last housed out of state and 30% in other California counties, and among those last housed in SF, the share there under one year rose from 6% (2019) to 32% (2024, 414 respondents). Cash aid handbook; HSH memo; HSH Point-in-Time survey dashboard, tab E, questions E1 and E2; Arrivals and residency

              Is the City funding children's programs?

              Yes. In the Mayor's FY26-27 budget, the Children's Baseline is funded at $293.6M against a $260.4M requirement, and DCYF, the City's children's department, grew $25.2M in the adopted budget ($23.0M in the proposal), mainly from Charter-required Student Success Fund and PEEF increases.

              More

              Ordinance 140-26; Controller, June 2026; Budget Book

              How many children live near the Tenderloin drug market?

              About 3,000 to 3,500 children live in the Tenderloin. Tenderloin Community Elementary is its only public elementary school. Safe Passage, a nonprofit with a budget of about $600,000, helped more than 1,000 children a day on average in 2023. Mission Local, Aug 2024 (press); AP, May 2024 (press)

              Tough questions, answered

              The City has its own deficit: about $450M in FY28-29 (preliminary), and FY27-28 balances with one-time money.

              True, and deficits claim savings first. Parts 3 and 4 touch the General Fund and have a real cost: $10.0M a year, and up to about $25.0M in FY27-28 and $26.8M in FY28-29 if the City would otherwise cut to the floor (est.). The City already uses Prop C to replace HSH General Fund support, so Part 3's swap competes with the deficit. The City Attorney should confirm our reading of Parts 1, 2, and 5.

              More

              True, and deficits claim savings first. On our reading, Parts 1 and 2 use money whose rules allow these uses if the services qualify. Part 5 is available only in a year the March deficit test is met or Excess ERAF (extra local property tax returned to the City) falls by half, and it adds money only when that year's appropriation is below the Charter amount. The City Attorney should confirm. Parts 3 and 4 touch the General Fund, at $10.0M a year and up to about $25.0M in FY27-28 and $26.8M in FY28-29 if the City would otherwise cut to the floor (est.), and have a real cost. Part 3's Prop C swap is a lever the City already uses for its own budget: in April 2026, the Mayor's Budget Director told Supervisors that unexpected Prop C money was replacing General Fund support for HSH. So this is a choice about priorities. In March 2026 the City projected $474.3M for FY27-28 and $741.7M for FY28-29. The budget signed in July closed FY27-28, leaning on one-time money. The Controller's June 2026 review of the Mayor's proposal put FY28-29 at about $450M, a preliminary figure from before the November ballot. Voters approved the Muni parcel tax and the regional transit measure, which that review assumed would pass (preliminary results). The December 2026 plan will reset it. Controller, June 2026; Mayor signs the budget, July 2026; Controller, March 2026; Charter text; Ordinance 140-26, PDF p. 13 ($458.3M of prior-year fund balance in FY27-28); 48 Hills, April 2026 (press); SF Elections, Nov 3, 2026 results (preliminary); regional transit measure results (KQED)

              SFUSD's spending doubled and scores didn't move, so why more money?

              That is a fair objection. Spending per student after inflation rose 106% from FY05-06 to FY24-25 (est.), and reading fell from 56.5% (2019) to 53.5% (2026). So the ask funds tutoring and attendance for students below standard, with a requested SFUSD match and quarterly public results. None of it goes to general operations.

              More

              Spending rose 35% from FY18-19 (est.). The money starts at the highest-poverty schools. SFUSD over 20 years State Current Expense of Education; State test results for SFUSD, 2026

              High-poverty schools already get more money.

              True, and much of it by design. SFUSD's 2026-27 plan gives schools over 75% low-income about $16,000 per projected student, against about $9,700 at schools under 40% (est.). School size matters too. That doesn't show whether the money works, so the ask requests quarterly results from SFUSD, including dollars spent by school.

              More

              Staff at average cost plus listed dollars, per projected student (est.): $9,701 at schools under 40% low-income, $9,978 at 40% to 60%, $11,367 at 60% to 75%, and $16,020 over 75%. Among schools of 200 to 799 students, it's $14,719 over 75% against $9,982 under 40%. About two-thirds of the gap comes from lines aimed at need or awarded by grant, such as Title I, Community Schools grants, and the Student Success Fund. Most of the rest is other staffing, which runs higher at smaller schools (est.). Schools under 200 students get about $20,600, and those with 800 or more about $9,300 (est.). Together, school size (measured as 1 over enrollment) and low-income share account for about 74% of the differences across the 99 schools with at least 50 projected students and a low-income figure (est.). Measuring size other ways gives 52% to 66%. The 2025-26 plan shows the same pattern, at about $13,600 against $9,400 (est.). The rise to $16,000 mostly reflects lower enrollment projections at those schools. On 2025-26 enrollment, the gap is about 54% in both plans. This is a March 2026 plan, subject to change, and not actual spending. It leaves out special education teachers and aides, other paraeducators, nurses, custodians, food, buses, substitutes other than dedicated core subs, and central offices. Average costs hide seniority, so the real dollar gap could be smaller. Look up any school in Our schools. SFUSD's 2026-27 school allocation files; 2025-26 files

              SFUSD balanced its budget and holds a large fund balance.

              True. The June 2026 budget is balanced for now. SFUSD started FY25-26 with $429M in fund balance, about half of it restricted. SFUSD says fund balance is one-time money. For FY28-29 it projects a gap of about $26M in unrestricted money. Because the balance is real, the ask requests an SFUSD match.

              More

              True. The June 2026 budget is balanced for now. SFUSD started FY25-26 with $429M in fund balance, about half of it restricted, and its June 2026 budget projects about $323M in its General Fund at the end of FY26-27, $236M of it restricted, plus a $111.5M reserve. SFUSD says fund balance is one-time money. For FY28-29 it projects a gap of about $26M in unrestricted money. It projects restricted funds running about $66M short in FY27-28 and about $53M short in FY28-29, leaving about $117M, in a projection that assumes voters renew QTEA. Because the balance is real, the ask requests an SFUSD match. SFUSD 2026-27 budget; Parents for Public Schools (secondary); SFUSD, Facts on the Budget; SF Standard, June 2026 (press)

              The State should pay for schools.

              In cash, the State pays about 46% of SFUSD's general fund in FY26-27, and its formulas and grants set about 70% of revenue (est.). San Francisco voters chose to add local money for schools. The ask is small and ties City money to attendance, which can also bring in State dollars slowly (est.).

              More

              In cash, the State pays about 46% of SFUSD's general fund in FY26-27, and its formulas and grants set about 70% of revenue (est.). San Francisco voters chose to add local money for schools. The ask is small and ties City money to attendance, which can also bring in State dollars slowly: nothing in year 1, about $0.2M to $0.4M in year 2, and up to $0.6M to $1.26M a year by year 4, from about $0.1M a year of outreach (est.). SFUSD 2026-27 budget, Form 01; State Budget Overview; SFUSD PEEF page (2004 and 2014 voter approvals); What more money would do

              The City already gives SFUSD about $168M a year.

              Largely true: the Charter requires about $167.9M in FY26-27 (est.), or about $225M with the City-levied Fair Wages for Educators parcel tax (est.), though SFUSD spent $21.7M of the $34.8M the Student Success Fund allocated to it in FY24-25. We found no causal study of the main set-asides, and the published evaluations track program metrics, so this ask adds quarterly public reporting.

              More

              Res. 471-25; Controller, June 2026; SFUSD PEEF page, reports (2022-23 evaluation); DCYF Student Success Fund evaluation RFP, Apr 2026

              The City already pays for reading tutoring, so why $10M more?

              True, for a smaller, lighter program. In 2025-26, Student Success Fund money helped expand short in-class reading tutoring to more than 2,700 students at 20 schools, at about $500 a student (press reports). Its 2026-27 funding is not confirmed. Part 3 would tutor 4,000 to 8,000 students below the reading or math standard, starting at schools over 75% low-income, at $1,200 to $2,500 a student (est.). In 2026, 10,812 tested students were below the reading standard and 12,583 below the math standard.

              More

              The strongest form of this objection: the City could grow tutoring inside the Student Success Fund without new money. That is our fallback for Part 3. It is not new money, it could reach only schools that meet the Fund's rules, and it exists only if the City does not freeze the Fund. Money set aside has also gone unspent. In FY24-25, SFUSD spent $21.7M of the $34.8M the Fund allocated to it, and the Fund's District Innovation and Workforce Pipeline line spent $2.8M of $7.25M. That is why the ask adds quarterly public reporting by school. The January 2026 expansion focused on grades 2 to 4; Part 3 counts students below the standard in tested grades 3 to 8 and 11. See What money can do. DCYF, Jan 20, 2026; Fund annual report, PDF p. 32

              SFUSD has empty seats and deferred closures.

              True. SFUSD had space for 14,000 more students across its campuses as of December 2025. It says no school is closing or merging for now. This ask doesn't settle that question. It funds specific programs for three years with public results, and it doesn't pay to keep empty seats open.

              More

              True. SFUSD had space for 14,000 more students across its campuses as of December 2025. It says no school is closing or merging for now, while a new assignment system is set to start in 2028-29, and it told KQED any closures or mergers would finish by fall 2030. This ask doesn't settle that question. It funds specific programs for three years with public results, and it doesn't pay to keep empty seats open. SF Standard, Dec 2025 (press); SFUSD Study Session 2 recap; KQED, May 2026 (press)

              Moving money will hurt people experiencing homelessness.

              It could. Part 3 moves $10.0M a year of HSH supportive housing costs from the General Fund to Prop C. No current program would be cut, but that's $10.0M a year less for new housing, renewals, and expected federal cuts, while family homelessness rose 15% from 2024 to 2026.

              More

              It could. Part 3 moves $10.0M a year of HSH supportive housing costs from the General Fund to Prop C, about 1.3% of HSH's budget and about 2.4% of its Prop C plan in FY27-28, the first year this could start. Tenants and providers should see no change from the swap itself, but those leases would then depend on a tax whose revenue swings from year to year. In FY27-28, HSH's Prop C plan already spends every dollar, so the $10.0M would replace planned new spending, such as new housing subsidies. Later years show room mainly because time-limited subsidies end, and HSH says unspent Prop C money could cover an expected federal gap of about $22M a year in supportive housing for three to four years. The ask leaves alone the $98M Prop C reserve the budget holds for federal and state funding risks. Part 3 is a three-year pilot, so the swap should end with it, or the City should say who pays after year three. Family homelessness rose 15% from 2024 to 2026, and HSH's adopted budget already falls from $832.7M to $780.3M in FY27-28, with a $36M shelter gap from FY28-29. HSH FY26-28; HSH, Jan 2026; HSH 2026 count; Ordinance 140-26, PDF p. 130 (HSH General Fund) and p. 264 (Sec. 32.2 reserve); HSH budget memo, February 2026, PDF pp. 2 and 5; HSH Prop C plan, June 2026, PDF pp. 10, 11, and 18; Who controls the money

              See the data: Arrivals and residency

              County cash aid requires 15 verified days in San Francisco (30 for PAES), and supportive housing is rationed by vulnerability: 10% to 20% of assessed adults qualify. Recent arrivals are rising: 41% of homeless residents in 2024 were last housed outside SF, mostly elsewhere in California.

              How the rules work: 15 verified days for cash aid, no entitlement to housing

              Cash aid

              15 continuous days of verified presence for General Assistance, CALM, and SSIP; 30 days for PAES. Since 2025, presence is verified through City records. Under Prop F, recipients with a substance use disorder must engage in treatment to keep county cash aid, with some exemptions, such as SSIP clients.

              Housing

              No entitlement. Supportive housing is rationed by a vulnerability score, and HSH projects 10% to 20% of assessed adults qualify. Shelter has no residency waiting period, though a bed may not be open.

              Sources: CAAP Eligibility Handbook, Oct 2026; CAAP Information Sheet; Treatment Pathway flyer; SF Examiner, Mar 2025 (press); HSH Definitions; HSH Prioritization Memo, Aug 2026; Mission Local, July 2026 (press); Shapiro v. Thompson; Saenz v. Roe.

              Who is arriving: the out-of-state share is 11%, the same as 2015; other California counties rose to 30%

              Show data table

              Where respondents were living when they became homeless, point-in-time survey. The 2019 report text gives 70% / 22% / 8%. 16% of respondents who had not always lived in SF said they came for homeless services or benefits (2024, 110 of 675). Source: HSH 2024 PIT page.

              Among those last housed in SF, the share there under one year rose from 6% (2019) to 17% (2022) to 32% (2024)

              Share who had lived in SF under one year (bars) and 10 or more years (dots) before most recently becoming homeless, of respondents housed in SF. The 2024 figure rests on 414 respondents. HSH's survey dashboard gives 13% and 46% for 2022, against 17% and 35% in the 2022 report, and warns that method changes limit comparisons across years. Sources: 2017, 2019, 2022, 2024 PIT reports.

              8.2% of 2025 overdose decedents had an address outside San Francisco

              625 accidental overdose deaths, 2025, by the decedent's verifiable address. "No fixed address" includes long-time residents who lost housing; it does not mean "from elsewhere." An SF address can be an SRO or supportive housing unit obtained after arriving. Jan to Sep 2026 (preliminary): 5.8% out of county, 22.6% no fixed address. Sources: OCME 2025; OCME Oct 2026.

              Homeward Bound served 430 households in FY24-25; 61 were confirmed housed at 90 days

              430Households served, FY24-25 (Journey Home 230, Problem Solving 197, CAAP Relocation 3)
              $454Travel cost per person ($187,328 total)
              215Were unsheltered before travel

              90-day follow-up

              Reuse is barred for two years (19 waivers granted). After January 2026 reforms, the City says more than 100 people were reconnected with family in three months. Sources: HSH Homeward Bound Report FY24-25; Mayor's release, May 12, 2026.

              Sources

              HSH Point-in-Time survey dashboard, tab E, question E3; E2.

              Sources

              Each answer links its own source. All sources and method.

              12 of 14 · Dates

              The Mayor's June 1, 2027 budget is the main decision, and budget work starts in December 2026.

              In our judgment, requests that are not in a department's submission by late February 2027 are much harder to add later.

              Firm: Jan 8, 2027 and Jun 1, 2027. Other City dates are expected, from prior cycles. The City has not published its 2027 calendar yet.

              See the data: Full decision calendar

              Most decisions that matter are made by June 1, 2027, when the Mayor proposes the budget; Board add-backs are set in committee in late June (June 25 in 2026), before the full Board's July votes, and are only about $28M to $41M a year citywide. The window runs from December 2026 department hearings to that date.

              Firm: Nov 3 and Nov 7, 2026 (past); Jan 8, 2027 (new terms start); Jun 1, 2027 (first business day of June); Jun 30, 2028. SFUSD's dates are from its published calendar. DPH's Jun 30, 2027 end date is its plan. Expected, from prior cycles: budget instructions (Dec 12 last time), Five-Year Plan (Dec 18 and 19 the last two times), department hearings (by Feb 14), submissions (Feb 23 last time), March update (Mar 31 last time), add-backs (Jun 25 last time), Board votes (Jul 21 and 28 last time, deadline by Aug 1). The City has not published its 2027 calendar yet.

              Sources

              Mayor's Budget Instructions, Dec 2025 (the prior cycle); SFUSD announcement, Oct 2026; Controller Revenue Letter, June 2026 (deficit); Controller March 2026 Update; DPH deck on the BHSA plan, Health Commission, Apr 2026; Mission Local, Oct 2026 (press); SF Standard, June 2026 (press); Mission Local, June 2026 (press) and Sup. Chan's July 2026 column (add-backs); SFUSD 2026-27 budget, PDF p. 160 (QTEA expires in 2028); SFUSD Board calendar; Board budget hearing calendar, 2026; Five-Year Plan, Dec 2024; Ballot measures, Nov 2026; SF Elections, Nov 3, 2026; State Prop 43; Welfare and Institutions Code 5963.02; KQED voter guide (three board seats; press); Charter Sec. 2.101 (Supervisors' terms start January 8); Sec. 13.101 (school board terms start the same day); Charter text, Sec. 16.131(d)(3) and (d)(6), PDF pp. 22-23; DHCS BHSA County Policy Manual, Ch. 3, Sec. A.3 (2027 update dates); Budget Book, PDF p. 13 (budget calendar and second-year revisit); Charter Sec. 9.103 (Board changes stay within the Mayor's fund totals); SPUR on 2008 Prop A (secondary; from July 1, 2008, for 20 years); SFUSD school reorganization page (final vote April 13, 2027).

              13 of 14 · Take action

              Write to the Mayor, the Board of Supervisors, and the school board before budget work starts in December.

              Write to the people who decide, in your own words or ours, and ask them where they stand.

              Mayor Daniel Lurie

              daniel.lurie@sfgov.org
              (415) 554-6141
              Mayor's page

              Board of Supervisors, all members

              board.of.supervisors@sfgov.org
              (415) 554-5184
              Board's contact page

              SFUSD Board of Education

              boardoffice@sfusd.edu
              (415) 241-6427
              How public comment works

              The next Regular meetings are December 1 and December 15, 2026, and January 12, 2027. There is no Regular meeting in November. Workshops limit comment to agenda items. Check SFUSD's calendar before you go.

              Each Supervisor, by district
              • Connie ChanDistrict 1Connie.Chan@sfgov.org(415) 554-7410
              • Stephen SherrillDistrict 2Stephen.Sherrill@sfgov.org(415) 554-7752
              • Danny SauterDistrict 3Danny.Sauter@sfgov.org(415) 554-7450
              • Alan WongDistrict 4Alan.Wong@sfgov.org(415) 554-7460
              • Bilal MahmoodDistrict 5Bilal.Mahmood@sfgov.org(415) 554-7630
              • Matt DorseyDistrict 6Matt.Dorsey@sfgov.org(415) 554-7970
              • Myrna MelgarDistrict 7Myrna.Melgar@sfgov.org(415) 554-6516
              • Rafael MandelmanDistrict 8, Board PresidentRafael.Mandelman@sfgov.org(415) 554-6968
              • Jackie FielderDistrict 9Jackie.Fielder@sfgov.org(415) 554-5144
              • Shamann WaltonDistrict 10Shamann.Walton@sfgov.org(415) 554-7670
              • Chyanne ChenDistrict 11Chyanne.Chen@sfgov.org(415) 554-6975

              Seats change on January 8, 2027. In preliminary results, Supervisors Sherrill (District 2), Wong (District 4), and Dorsey (District 6) won re-election. Gary McCoy will represent District 8 and Theo Ellington District 10, replacing Supervisors Mandelman and Walton. Supervisor Chan conceded her race for Congress, so she stays in District 1. The new Board elects its President on January 8 (SF Elections, November 10 count). Names and phones checked on sf.gov on November 1, 2026. Email addresses are from the Clerk of the Board's roster, effective April 6, 2026. The City's own pages now list office inboxes: ChanStaff, SherrillStaff, SauterStaff, WongStaff, MahmoodStaff, DorseyStaff, MelgarStaff, MandelmanStaff, FielderStaff, WaltonStaff, ChenStaff, all @sfgov.org.

              A short message for the Mayor and Supervisors

              Dear Mayor Lurie and Supervisors, I live in San Francisco, and I care about our public schools. Please put students first in next year's budget (FY27-28). Fund $15.5M a year for school-based prevention, early intervention, tutoring, and attendance. Aim the tutoring and attendance money at students below the standard, starting at our highest-poverty schools, where about 1 in 4 students meets the reading standard. Keep the Student Success Fund on schedule, and if the City freezes it, use its reserve, last reported at $18.1M, for students as the Charter allows. Please ask SFUSD to report results every quarter. The Mayor's budget is due June 1, 2027. Please reply and tell me where you stand. Thank you, [Your name] [Your neighborhood]

              A short message for the school board

              Dear Commissioners, I live in San Francisco, and I care about our public schools. Please commit SFUSD to a match for tutoring and attendance for students below the standard, starting at our highest-poverty schools, and to reporting results every quarter. Your budget hearing and vote are expected June 8 and June 22, 2027. Please reply and tell me where you stand. Thank you, [Your name] [Your neighborhood]
              14 of 14 · Who we are

              Who we are.

              [Who's behind this]
              [Contact]

              How this was made

              We built this from public City, State, and SFUSD documents, with research and drafting help from AI tools. Figures marked (est.) are our own calculations or judgments. No outside expert, City office, or SFUSD has reviewed this site. Advocacy analysis by a group with a stated position. Not an official City or SFUSD product.

              See the data: Sources and method

              Figures on this site trace to public documents. Calculations beyond simple sums and shares are marked "est.", and the method notes explain the main ones.

              How this was made

              We built this from public City, State, and SFUSD documents, with research and drafting help from AI tools. Figures link to their sources in each section. Figures marked (est.) are our own calculations or judgments, and the movable-dollar ranges are judgments. Independent data sites are labeled secondary.

              We rechecked figures against sources in November 2026, using separate AI review passes that we ran ourselves. No outside expert, City office, or SFUSD has reviewed this site. Items we couldn't verify are listed in Method.

              Advocacy analysis by a group with a stated position. Not an official City or SFUSD product.

              Method

              • Students below the State standard = students with scores minus those the State counts as Proficient or Advanced, its 2026 names for meeting or exceeding the standard. SFUSD's district row, which leaves out charter schools, grades 3 to 8 and 11, spring 2026. Math: 12,583 of 23,218 (54.2%). Reading: 10,812 of 23,251 (46.5%). The dots show each group's share meeting the standard, one dot for each student in ten. Groups with under 300 students tested aren't charted: Native Hawaiian or Pacific Islander (141 tested in math, 12.8%; 144 in reading, 22.9%) and American Indian or Alaska Native (31, 19.4%; 33, 39.4%). Sources: State test results, SF file, 2026; SFUSD on the State's results site, 2025-26.
              • SFUSD next to California = SFUSD's district row set against all non-charter public schools in California, for the same student groups, tests, grades, and year, as the State's results site showed them on November 16, 2026. Both sides leave out charter schools. On that basis SFUSD is 7.2 points above California in math and 4.2 in reading for all students. Four of six racial and ethnic groups are below the same groups statewide: Asian (4.4 points in math, 3.9 in reading), Black or African American (9.4, 12.0), Filipino (19.1, 15.2), and Hispanic or Latino (8.4, 11.7). Two are above: Two or more races (3.0, 3.2) and White (13.6, 13.8). These are group averages. They don't show why, and the mix of students inside each group differs between San Francisco and the State. Sources: State test results, SF file, 2026; California's non-charter schools on the State's results site, 2025-26.
              • Part 3 next to the students below the standard = Part 3's $9.9M for tutoring ÷ $2,500 a student = about 4,000 seats, and ÷ $1,200 = about 8,000 (est.). Against 12,583 tested students below the math standard, that is about 32% to 64% (est.), drawn as 32 and 64 of 100 squares. Against the 10,812 below the reading standard, it is about 37% to 74% (est.). These are scale comparisons, not counts of who is served: Part 3 serves students below the reading or math standard, some students are below both, and grades without State tests aren't counted. Insights uses a narrower base, the 7,846 low-income students below the reading standard, all of whom $10.0M could reach at $1,200 each (est.). Sources: State test results, SF file, 2026; the ask, Part 3.
              • Math by income and school poverty = for each of the 68 district-run elementary and K-8 schools with at least 100 students (Census Day 2025-26), students meeting the math standard and students tested in 2026, for low-income students and for everyone else (all students minus low-income), summed within bands of the school's low-income share. Counts come from the State's percentages and tested counts. Four smaller schools are left out: Carver (97 students), Malcolm X Academy (91), Mission Education Center (75), and Edwin and Anita Lee Newcomer (15). K-5 schools alone: 46.7% and 73.9% at schools under 40% low-income, 25.3% and 38.7% over 75%. At schools over 75% low-income, 2,714 of 3,584 tested students were below the math standard (76%). Across the 68 schools, 6,272 were below, and 3,558 of them (57%) attended schools at or under 75% low-income (est.). Sources: State test results, SF file, 2026; State census day, 2025-26.
              • Students per school social worker = 50,028 SFUSD students in 2025-26 (not counting charter or county schools) ÷ 99 school-site social workers in FY25-26 = about 505, or about 1 per 500 (est.). The two counts come from different sources. The School Social Work Association of America recommends no more than 250 students per social worker. Sources: SFUSD Facts at a Glance, 2025-26; SF Standard, June 26, 2026 (press); SSWAA ratio resolution.
              • City money per student = PEEF to SFUSD ($107.6M) + Student Success Fund ($45.0M) + Public Education Services Baseline ($15.3M), divided by about 50,000 district-run students. FY26-27. The State's Census Day 2025-26 count of 48,472 non-charter students would give about $3,460 (est.). About 50,000 is SFUSD's 50,028 for 2025-26. Sources: SF Controller, FY26-27 and FY27-28 Revenue Letter, Exhibit 4-1; Ordinance 140-26; SFUSD Facts at a Glance; CDE Census Day enrollment 2025-26.
              • City money per HSH client = HSH FY24-25 actual spending ($629.8M) × its local share, divided by the people it served. Local share: about 80% in FY24-25, from the Controller's spending data by fund. Grant funds were $125.7M of the $629.8M, so City money (General Fund and Our City, Our Home) was about $504.0M. People: the low end divides by all 43,148 people, meaning the 26,634 HSH served while unhoused in 2024 plus the 16,514 in its housing programs, assuming no overlap. The high end divides by the 26,634 only. The report doesn't say how far the two groups overlap. City money, FY24-25 actual: $504.0M ÷ 43,148 = about $11,700; $504.0M ÷ 26,634 = about $18,900 (est.). All funds, FY24-25 actual: $629.8M ÷ 43,148 = about $14,600; $629.8M ÷ 26,634 = about $23,650 (est.). On the FY26-27 budget ($832.7M), with the same 2024 counts and HSH's budgeted local shares (83% in FY22-23, about 90% in FY26-28): City money $832.7M × 83% ÷ 43,148 = about $16,000, and $832.7M × 90% ÷ 26,634 = about $28,100; all funds $832.7M ÷ 43,148 = about $19,300, and $832.7M ÷ 26,634 = about $31,300 (est.). The 83.0% is FY22-23: $285M General Fund + $233M Prop C + $40M other local = $558M of $672M. The 89.9% is FY26-28: $813.9M Prop C + $525.5M General Fund + $110.4M other local = $1,449.8M of $1,613.0M. Sources: Controller spending data (DataSF); HSH budget and funding overview (May 2023), slide 4; HSH Budget FY26-28; Controller 2025 Homelessness Needs Assessment, PDF pp. 7, 20, and 34.
              • Local money outside the State formula, per student = local SFUSD and County Office revenue outside the formula in FY26-27 ($361.6M: City set-asides, parcel taxes, other local), divided by about 50,000 students = about $7,230 (est.). This treats voter-approved school parcel taxes the way Prop C is treated on the HSH side. It leaves out the about $329M of local property tax inside the State formula; with it, about $13,800 (est.). On 48,472 students, $7,230 would be about $7,460 (est.). The $329M is $316.5M + $43.1M less $30.5M on Form 01. Sources: CDE Budget Overview for Parents, FY26-27; SFUSD 2026-27 Recommended Budget, Form 01.
              • Same-year check. On City set-asides alone, using one year on both sides: FY26-27 budgets give about $16,000 to $28,100 per person HSH served, against about $3,360 per student; FY24-25 gives about $11,700 to $18,900, against about $2,900 per student ($145.0M) (est.). The $145.0M is $96.3M of PEEF to SFUSD + $35.0M Student Success Fund + $13.7M Public Education Services Baseline. Source: SF Controller, March 2026 Update, Appendix E, p. 14.
              • $1.53B = HSH FY26-27 adopted budget ($832.7M) plus DPH's Behavioral Health division in the adopted budget ($698.2M) = $1,531M (est.). DPH's broader behavioral health total in its April plan, including ZSFG and Laguna Honda, is $795.6M, giving about $1.63B. Both include mental health, not only addiction. Sources: Ordinance 140-26, pp. 130 and 178; DPH deck on the BHSA plan, slide 12.
              • DPH behavioral health vendor payments = all FY payments in the Controller's vouchers data to 26 vendors whose DPH contracts are behavioral health by title, plus behavioral-health-titled contracts at other DPH vendors. Excludes the UC affiliation agreement, nursing facility contracts, and IT vendors. An estimate; HSH figures are exact department filters.
              • Overdose rates = CDC provisional deaths by county of residence (12 months ending December) divided by Census population (2024 population used for 2025). Crude, not age-adjusted.
              • Real dollars = nominal values times the ratio of the FY25-26 average SF-area CPI-U (363.78) to the fiscal-year average for that year. Real spending per student: $13,393 (FY05-06), $14,310 (FY13-14), $20,435 (FY18-19), $27,642 (FY24-25), up 106% since FY05-06 and 35% since FY18-19 (est.). Of the $14,249 gain, about 43% came from FY13-14 to FY18-19 and about 51% from FY18-19 to FY24-25 (est.). Homelessness went from $84.2M (FY06-07) to $640.5M (FY24-25), or $143.8M to $661.5M after inflation, 4.6 times. Substance use treatment went from $61.1M to $120.9M, or $104.4M to $124.9M, about 1.2 times, on definitions that differ. Combined, $145.3M to $761.4M, or $248.2M to $786.4M after inflation, about 3.2 times (est.). Sources: BLS CPI-U, San Francisco-Oakland-Hayward; Controller spending data (DataSF); DPH Treatment on Demand Report FY24-25, p. 9.
              • Methadone fill rate = people served in the year divided by the point-in-time capacity DPH reported (contracted through FY22-23; DPH's FY23-24 report calls it licensed). An upper bound: daily census is lower than annual people served. Sources: DPH Treatment on Demand Report FY21-22; DPH Treatment on Demand Report FY22-23; DataSF, DPH Substance Use Services (ubf6-e57x).
              • Population counts from survey percentages multiply the share by the matching population (for example, NSDUH 1.83% opioid use disorder times about 739,800 residents 12 and older). The 739,800 is the 2021 to 2023 average: total population, less those under 10, less 40% of ages 10 to 14 (est.). NSDUH 1.83% gives about 13,500, 12.04% gives about 89,100, and 21.64% gives about 160,100 (est.). Homeless adults who regularly use drugs: 35% to 37.1% × 26,634 = about 9,300 to 9,900 (est.). Homeless people with a substance use disorder diagnosis: 84% × 16,671 = about 14,000 (est.). The 15,000 to 17,000 who use fentanyl regularly, range 7,000 to 26,000, come from a secondary site's planning model (est.). Sources: SAMHSA NSDUH 2021-2023 substate estimates; Census county population by age, Vintage 2024; UCSF BHHI Behavioral Health Report; Assaf et al., JAMA 2025; Controller 2025 Homelessness Needs Assessment.
              • Substance use treatment per person = DPH specialty substance use budget ($120.9M, FY24-25) divided by 17,029 people treated, about $7,100 (est.). The City-money figure uses the General Fund plus Prop C share: $39.86M + $10.99M = $50.8M, about $3,000 per person (est.). Source: DPH Treatment on Demand Report FY24-25, pp. 4, 8, 9.
              • Budget vs. actual. HSH spent 75% of its FY24-25 budget on the Controller's net basis ($629.8M of $834.6M), the basis used for headlines here. Department gross gives $640.5M of $846.3M, or 76%; the budget chart in City homelessness spending uses that basis and says so. Sources: Controller spending data (DataSF); Controller budget data (DataSF).
              • Chronic absence. District-run schools: 14.2% (2018-19), 23.9% (2024-25), 27.8% (2025-26). Counting SFUSD-authorized charters: 18.8%, 28.8%, 32.7%. Five Keys and other alternative-status charters account for the jump from 27.8% to 32.7%. Without alternative-status schools, the State's 2025-26 rate is 27.2%, with district-run schools at 27.3% and SFUSD-authorized charters at 25.8%. Ed-Data's district rate is the second series. Each mention on this site says which it is.
              • Teachers. SFUSD's own count: 3,387 (2022-23) to 2,947 (2025-26). Ed-Data's full-time-equivalent series: 3,348 (FY09-10) to 2,857 (FY25-26), with breaks. The two are not comparable, so this site quotes no single percentage change across them. Ed-Data's school charts (chart 180), summed by us: 2,274 at district-run schools in 2018-19 and 2,566 in 2025-26; charter schools 338 and 306. The gap to the district figure in 2018-19 is 3,774.8 - 2,273.7 - 338.4, about 1,160 (est.).
              • School budgets. SFUSD's 2026-27 school allocation files, one per school for 102 schools, marked March allocations, revised March 30 to April 14, 2026, and subject to change. They list staff positions and some dollars, not a total. These files differ from the March 5 versions on SFUSD's Get Involved page. For example, ESSA school improvement is $600,000 at S.F. International, Downtown, and Wells here, and $150,000 there. SFUSD should confirm whether that's one year's money. Enrollment-based dollars here still appear to use the earlier enrollment figures. Our total: certificated positions x $161,000 (our FY26-27 teacher estimate), classified positions (clerks, security aides, transitional kindergarten paraeducators, noon monitors, community health outreach workers) x the midpoint of $58,000 to $87,000, plus listed dollars. The 73.7 positions of Executive Cabinet-approved funding, 33 dedicated core subs, and 12 JROTC instructors count as certificated, about $19M in all, though the files don't give their job types or costs. Principals cost more, and the classified range assumes 6 to 7 hours a day, so those two run low by about 1% each. Items marked "to be determined" count as $0, including arts Prop 28 dollars and AmeriCorps interns at every school and the Student Success Fund at seven. Title I is listed at 80% of the projected allocation. Literacy grants are listed as anticipated carryover, and PITCH amounts as exact award to be determined. Low-income groups use 2025-26 shares, so Mission Bay isn't in them. Per student uses each file's projected enrollment (48,066, including special day class students, whose teachers aren't in the total), and we leave it blank under 50 projected students (Lee Newcomer and Mission Education Center). Those two schools still count in group totals and the $547M. It leaves out special education teachers and aides, other paraeducators, nurses, custodians, food, buses, substitutes other than dedicated core subs, and central offices. Average costs hide seniority, so the dollar gap by need could be smaller than shown. Position counts and listed dollars don't have this problem. This is a plan, not actual spending. Position categories changed from the 2025-26 files, marked preliminary in February 2025, so year-to-year comparisons need care. Lowell's $2,467,944 for Advanced Placement prep is 37% of that line's total, and its 2025-26 file shows a similar $2,520,744. SFUSD's file ties it to UESF contract section 7.2.11. SFUSD should confirm what it covers.
              • School by school. One row per district-run school on CDE's 2025-26 Census Day list, plus four County Office schools. Supervisor district is where a school stands, from City address records on the 2022 district map. We checked it against State coordinates. 100 of 101 agreed. Roosevelt Middle sits on the line between Districts 1 and 2, and the City's record puts it in District 2. Low-income = the State's socioeconomically disadvantaged share of enrollment. % full = 2025-26 enrollment ÷ 2022-23 program capacity. Correlations and medians are over the 99 to 101 district-run K-12 schools with data (est.). Reading on 2026 tests against low-income share, as charted: slope -0.84, r = -0.81, so low-income share accounts for about 66% of the variation between schools (r squared, est.), as in Insights. On 2025 tests it was about 58%. Suppressed scores (under 11 tested) and homeless counts under 11 are left blank. Sources: CDE Census Day enrollment 2025-26; CAASPP research files (2019, 2025, 2026); SFUSD 2023 Facilities Master Plan.
              • Control map funnel = HSH General Fund ($252.1M, adopted) + DPH substance use General Fund (about $39.9M) + DPH behavioral health non-substance-use General Fund contracts (about $90M, a June 2024 base before later cuts) + Jail Health General Fund share (about $43.8M: $58.0M less about $14.2M of Medi-Cal and other revenue) + RESET (about $9.0M) + Fire teams General Fund share (about $6.9M: about $18.0M less the $11.1M Prop C work order) = about $442M (est., mixed years). Legal outer bound removes Jail Health and caps HSH cuts at Prop C baseline headroom ($159.6M to $173.8M, on the Mayor's proposed $374.6M): about $305M to $320M. Realistic = the sum of row estimates ($7M to $58M), which are our judgment, not calculations. Restricted channels = Prop C families ($5M to $20M) + BHSA youth ($1M to $5M) + HHAP youth ($0 to $2M, one-time grants spread over their terms) + block grant prevention ($0 to $2.4M) = about $6M to $29M. All channels = realistic + restricted = about $13M to $87M (est.). Of that, $5M to $22M is homelessness money only: Prop C families plus HHAP youth. The ask's three channels = HSH General Fund ($5M to $30M) + opioid settlement ($2M to $8M) + BHSA youth ($1M to $5M) = $8M to $43M (est.). HHAP youth: 10% of Round 7's $45.6M is about $4.6M; 10% of Round 6's $39.9M is about $4.0M; spread over their terms, at most about $1.8M a year (est.). BHSA youth: $32.3M × 51% × 51% = about $8.4M, on DPH's April plan (est.). Block grant: $9.5M × 25% = about $2.4M (est.). Sources: Ordinance 140-26, pp. 17 and 178; DPH Treatment on Demand Report FY24-25, p. 8; DPH budget hearing attachment (contract split), p. 1; DPH budget submission, FY27 and FY28, p. 11 (Jail Health revenue of $7,480,000 + $5,232,071 + $967,410 + $477,986); Sheriff Budget Submission FY27 and FY28; SFFD Budget Book FY25 and FY26, pp. 105, 108, 113; Controller Budget (DataSF xdgd-c79v) ($11,095,562: DPH, Our City Our Home fund, object GF-Fire); SF Controller, FY26-27 and FY27-28 Revenue Letter, p. 39; SF Controller, March 2026 Update, Appendix E, p. 14; HCD HHAP Round 7 preliminary allocations; HSH HHAP Round 6 deck, pp. 4 and 8; JBAY HHAP fact sheet, p. 2; DPH deck on the BHSA plan, slide 16; DHCS Substance Use Block Grant, Exhibit A.
              • Part 3's funding swap. Share of HSH's budget = $10.0M ÷ $780.3M (FY27-28 adopted) = 1.3%. Share of HSH's Prop C plan = $10.0M ÷ $421.68M of planned FY27-28 uses = 2.4%. Planned sources that year are also $421.68M. FY28-29 and FY29-30 show $66.97M and $77.27M of room, mostly because time-limited items end. Precedent: HSH's February 2026 budget memo met a $4M General Fund target by moving the Raphael House family shelter to Prop C and the Sharon Hotel to opioid settlement funds. The same memo expects a federal gap of about $22M a year in supportive housing. Reserve: $98M (Ordinance 140-26, Sec. 32.2). Press: 48 Hills (April 30, 2026) reports the Mayor's Budget Director telling Supervisors that unexpected Prop C money was replacing HSH General Fund support. HSH, the Controller, and the City Attorney have not reviewed this reading. Sources: HSH budget memo, February 2026, PDF pp. 2 and 5; HSH Prop C plan, June 2026, PDF pp. 10, 11, and 18; Ordinance 140-26, PDF p. 264; 48 Hills, April 30, 2026.
              • City budget, 20 years. Actual = Controller Spending and Revenue (spending rows, related government units excluded); budget = Controller Budget dataset (from FY09-10). Both net of the Controller's citywide transfer adjustments; organization groups net, departments gross (Appropriation Ordinance convention). Real uses the SF CPI-U method above. Per resident uses Census July 1 population for the fiscal year's first year.
              • Share reaching SFUSD = PEEF two-thirds + Student Success Fund appropriation + Public Education Services Baseline, divided by the City all-funds total (FY06-07 to FY16-17: DCYF's recorded PEEF payments, actual, which lag and understate Charter amounts). Homelessness = predecessor programs, then HSH. Behavioral health = DPH community behavioral health programs to FY16-17, then the DPH Behavioral Health division. Sources: Controller Spending and Revenue (DataSF bpnb-jwfb); Controller Budget (DataSF xdgd-c79v); SF Controller, FY24-25 and FY25-26 Revenue Letter (June 2024); SF Controller, March 2026 Update; SF Controller, FY26-27 and FY27-28 Revenue Letter; Ordinance 140-26.
              • What it could buy = amount divided by unit cost. Benefits are 48.8% of salaries ($359.0M ÷ $735.2M, SFUSD 2026-27 Recommended Budget, Form MYP). Teacher about $161,000: FY24-25 average salary of $103,472 (Ed-Data) × 1.0485 schedule growth (UESF B8, step 10) × 1.488 (est.). Social worker about $152,000: $102,302 at step 5 on the 2026-27 schedule (UESF B9) × 1.488; about $169,000 at step 8, $113,504 (est.). Paraeducator about $58,000 to $87,000: $35.72 to $45.69 an hour (UESF Schedule 935) × 6 to 7 hours × 182 days × 1.488; the hours and days are our assumption (est.). Tutoring $2,500 (low count) to $1,200 (high count) per student, midpoint $1,850; the range is from EdSurge, a press summary. After-school seat $2,750 (CDE ELO-P rates, FY25-26, Rate 1). Illustrative only.
              • Study table costs are our estimates where the study gives none. Small classes: teacher about $161,000 ÷ 15 students, less ÷ 22 students = about $3,400. Counselor: about $152,000 ÷ 350 to 600 students = $250 to $450. Community schools: see Community schools cost.
              • Impact calculator. Even split: amount ÷ about 50,000 students; test scores +0.0316 SD per $1,000 in 2018 dollars sustained four years (Jackson & Mackevicius), so $1,000 today counts as about $760 (BLS CPI-U: 251.107 in 2018 and 336.940 in Sep 2026 gives $745; SF-area 285.550 in 2018 and 370.130 in Aug 2026 gives $771); low end one-third (Reber & Goodman); share meeting standard assumes normal scores and a 46% base; college +2.8 pp and graduation +1.92 pp per $1,000 × about 3,760 per graduating class (15,048 grade 9-12 students ÷ 4). Targeted: 2,996 homeless students at district-run schools (3,624 counting charters) × $1,200 to $2,500; 0.05 to 0.15 SD = +1.4 to +4.3 pp math (19.1% base) and +1.6 to +5.1 pp reading (25.8% base); +0.4 to +1.4 attendance days per tutored student, from D.C.'s preliminary result (6.9% less likely to be absent on days with a session) at three sessions a week (the release gives 2.3 and 5.2 days if sessions ran every school day; at three sessions a week for 36 weeks, about 60% of days, that is about 1.4 and about 3.1; the +0.4 low end is our judgment); mailers $6.60 × the households of about 13,700 students, 0.5 to 1.1 days each; about $83.50 State formula money per student-day ($677.5M ÷ 45,086 ÷ 180). The remainder is spread evenly. The $9.9M card is Part 3's $10.0M less about $0.1M for attendance outreach. $9.9M ÷ $2,500 = 3,960 students and $9.9M ÷ $1,200 = 8,250, shown as about 4,000 to 8,000 (est.). The first group, about 3,600 students, is about 36% of the 10,139 students at the 35 open schools over 75% low-income. The calculator uses the 2,996 homeless students at district-run schools. They are different groups. Sources: Jackson & Mackevicius, AEJ: Applied Economics 16(1), 2024; Jackson and Mackevicius, NBER w28517; NEPC Review of Reber & Goodman; SFUSD Facts at a Glance (50,028 students; 15,048 high school students ÷ 4 = about 3,760); State test results 2025 (46% math); CDE Census Day enrollment 2025-26 (2,996; 3,624; 10,139); Ed-Data, test results by student group (19.1%; 25.8%); National Student Support Accelerator on D.C. tutoring and attendance; Rogers and Feller supplement ($6.60); CDE chronic absenteeism 2025-26 (13,698 of 49,224); Ed-Data, LCFF revenue ($677.5M); CDE Current Expense of Education (45,086).
              • Cents per City dollar = City money to SFUSD ($167.9M, est.) divided by the all-funds budget ($16.85B), about 1 cent, or by the General Fund budget ($7.60B), about 2.2 cents (est.). FY26-27. With the City-levied Fair Wages for Educators parcel tax ($57.3M), $225.2M is about 1.3 cents (est.). On adopted budget lines ($116.3M PEEF + $16.5M PEEF baseline + $45.0M Student Success Fund = $177.8M), about 1.1 cents (est.). Sources: SF Controller, FY26-27 and FY27-28 Revenue Letter, Exhibit 4-1; Ordinance 140-26, pp. 13, 67, 68.
              • City money in homelessness and substance use = the HSH FY26-27 budget ($832.7M) times its local share of 83% to about 90%, plus the General Fund and Prop C share of specialty treatment ($50.8M, FY24-25) = about $742M to $799M (est., mixed years). Sources: Ordinance 140-26, p. 130; HSH Budget FY26-28; HSH budget and funding overview (May 2023); DPH Treatment on Demand Report FY24-25, pp. 8-9.
              • Percentile points = the normal curve at 0.03, 0.05, and 0.15 standard deviations for a student at the 50th percentile: about 1, 2, and 6 points (est.). Students at the 20th percentile gain about 1.4 to 4.5 points from 0.05 to 0.15. The 0.05 to 0.15 range is our judgment, set from the Personalized Learning Initiative (0.06 to 0.09 pooled, sites up to 0.16) and Nashville (0.04 to 0.09 in reading, none in math).
              • Attendance outreach = mailers at $6.60 (2014-15 dollars, about $9 today) for the households of about 13,700 chronically absent district-run students, about $0.1M a year, returning 0.5 to 1.1 days each at about $83.50 of State formula money per student-day. At full effect that is $0.6M to $1.26M a year, but SFUSD's three-year funding average spreads it: $0 in year 1, one-third in year 2, two-thirds in year 3 (est.). The 1.1 days is the study's two best letters (Rogers and Feller supplement, Table S13). The 0.5 is our low-end judgment, about half that; the study's Latino estimate was about half its average, though the authors found no significant group differences. The three-year average runs 451 to 890 students above projected attendance (funded 45,242, 45,077, 44,792; actual 44,791, 44,344, 43,902). At the 35 open schools over 75% low-income: about 4,635 chronically absent students (the State's count) × $6.60 = about $31,000, and × 1.1 days × $83.50 = about $0.4M a year (est.). Sources: SFUSD 2026-27 Recommended Budget, PDF p. 5; Education Code 42238.05 (public mirror); CDE chronic absenteeism 2025-26; CDE Census Day enrollment 2025-26; Rogers and Feller supplement, Tables 29 and S13; Ed-Data, LCFF revenue; CDE Current Expense of Education.
              • Homelessness and substance use, FY26-27, all funds = HSH's adopted budget ($832.7M) plus specialty treatment. The low end uses $120.9M, DPH's last published figure (FY24-25). The high end uses about $133M, which is the SF Funding Explorer's FY26-27 total ($965.5M, a secondary site's extrapolation) less $832.7M. About $954M to $966M (est., mixed years). Sources: Ordinance 140-26, p. 130; DPH Treatment on Demand Report FY24-25, p. 9; SF Funding Explorer.
              • Housing and shelter share = HSH's two-year housing budget ($909.0M) plus shelter and crisis ($377.3M), $1,286.3M of $1,613.0M, or 79.7% (HSH Budget FY26-28). 79.7% of $832.7M is about $664M, about 69% of $954M to $966M (est.).
              • Substance use treatment series = DPH's "Substance Abuse - Community Care" program (code DMS) in Controller spending data, before transfer adjustments, FY06-07 to FY16-17; then DPH's specialty budgets from Treatment on Demand reports, FY20-21 to FY24-25. The two definitions differ, and DPH says some programs moved in from its mental health budget.
              • Actual above budget. Actual spending ran 5% to 25% above the adopted budget each year. We read the gap as mostly multi-year capital and bond spending. In FY24-25, about $2.6B of actual spending sits in a fund category ("Unspecified") with no budget (about $2.5B of it in Public Works, Transportation & Commerce), led by Airport capital outlay ($733M), PUC non-personnel ($688M), and PUC debt service ($432M). This is our reading of Controller data. No City document states it.
              • Reading at the highest-poverty schools = tested students meeting the standard at the 35 open schools over 75% low-income: 25.4% (2025) and 25.5% (2026), about 1 in 4 (est.). At the typical such school, the median, it is 22.0% and 19.5%, about 1 in 5. State test result files and CDE enrollment.
              • SF's own opioid settlements. Walgreens about $230M + Allergan and Teva $54M + Endo $10M = about $294M. National settlements: CVS about $11M + Walmart $6M + Johnson & Johnson and the distributors $45M = about $62M. $294M of $356M is about 83%, or about four-fifths (est.). These are totals announced by May 2023, not cash received. Source: City Attorney, Walgreens settlement release.
              • Parts 4 and 5. FY27-28: $60.0M budgeted less the Charter's $35M floor = $25.0M; less a $45.0M freeze = $15.0M. FY28-29: $60.0M × 1.03 = $61.8M under the Charter's 3% cap; less $35M = $26.8M; less a $60.0M freeze = $1.8M. On the Controller's $62.9M for FY28-29: $27.9M and $2.9M (est.). Part 5's room is the gap between the Charter amount and the appropriation: about $15.0M in FY27-28 and about $2M to $3M in FY28-29 after a freeze. Sources: Ordinance 140-26, p. 67 ($45.0M and $60.0M); Charter Sec. 16.131(d)(2), (d)(3), (d)(6); SF Controller, March 2026 Update, Appendix E, p. 14 ($62.9M); Controller Nine-Month Report, FY25-26, p. 29 ($18.1M).
              • State and local shares of SFUSD money. General fund revenue in FY26-27 is $1,364.5M. The State formula is $713.2M, of which about $329.0M is local property tax ($316.5M + $43.1M less $30.5M to charters). State cash = State aid $375.1M + Education Protection Account $9.0M + other State revenue $243.3M = $627.4M, about 46% (est.). All local = $329.0M + $360.0M of other local money = $689.0M, about 50% (est.). The 70% and 26% use the State's Budget Overview, which includes the County Office. Sources: SFUSD 2026-27 Recommended Budget, Form 01 and Form MYP, PDF pp. 19, 20, 134; CDE Budget Overview for Parents, FY26-27.
              • Substance use, wider tally. $123M = specialty treatment ($120.9M) + Fire's overdose team ($1.7M) + its Tenderloin overdose project ($0.5M). Up to about $215M adds partly related costs at full value: Fire's crisis team ($12.5M), EMS-6 ($3.0M), Jail Health ($55.4M, FY25-26), and a District Attorney share ($20.9M: 22.5% of $93.0M, using a share from 2019) (est., mixed years). The 22.5% is the Budget and Legislative Analyst's 2019 midpoint for drug charges in District Attorney caseloads, and it includes people charged with selling. $215M ÷ 17,029 people = about $12,600 (est.). Sources: DPH Treatment on Demand Report FY24-25; SFFD Budget Book FY25 and FY26; DPH Jail Health deck, July 2025; Controller spending data (DataSF); Budget and Legislative Analyst memo, April 2019.
              • All funds per person. Student: about $27,000, from $26,762 per unit of attendance in FY24-25 to about $27,200 per student in FY26-27 ($1,361.8M ÷ about 50,000) (est.). HSH client: $629.8M ÷ 43,148 to ÷ 26,634 people = about $14,600 to $23,650 (est.). Sources: CDE Current Expense of Education; SFUSD 2026-27 Recommended Budget, Form MYP; Controller spending data (DataSF); Controller 2025 Homelessness Needs Assessment.
              • Homelessness and behavioral health totals. About $1.29B = HSH actual spending in FY24-25 ($629.8M) + DPH's Behavioral Health division budget in FY25-26 ($656.1M) (est., mixed years). $0.75B = $629.8M + $120.9M of specialty treatment. $0.97B = HSH's $846.3M budget + $120.9M. Outside providers: HSH vendors $536.0M + DPH behavioral health vendors about $392.1M = about $928M in FY24-25; $554.8M + about $404.8M = about $960M in FY25-26, so $0.93B to $0.96B. Nonprofits: $495.0M + about $306.3M = about $801M; $527.2M + about $337.2M = about $864M, so $0.80B to $0.86B (est.). Sources: Controller spending data (DataSF); Ordinance 140-26, p. 178; DPH Treatment on Demand Report FY24-25; Controller Vendor Payments (Vouchers), DataSF n9pm-xkyq.
              • School counts and ZIP shares. Non-charter enrollment fell from 52,947 (FY09-10) to 48,472 (2025-26), about 4,500 (est.). ZIP figures are ACS 2020-2024 estimates via Census Reporter: public K-12 students from table B14002 and poor children ages 5 to 17 from table B17001, each ZIP as a share of the San Francisco total. 42% and 54% add 94112, 94110, 94124, and 94134 (est.). ZIP shares pool public and private K-12 students across ZIPs: 15,680 of 19,872 in the 9 ZIPs with child poverty of 14% or more, about 79%; 10,526 of 20,654 in the 9 ZIPs under 4%, about 51% (est.). Survey margins are large. Sources: Ed-Data, enrollment; CDE Census Day enrollment 2025-26; ACS 2020-2024 by ZIP, via Census Reporter.
              • Overdose deaths by housing status. Rate gap: 143 deaths with no fixed address in 2025 ÷ 8,323 (2024 count) or ÷ 26,634 (HSH clients in 2024) = about 1,720 or about 540 per 100,000. 415 deaths with a fixed San Francisco address ÷ about 819,000 other residents = about 51 per 100,000. That is about 10 to 35 times (est.). Homeless or in subsidized rooms: 23% with no fixed address + about 18% to 26% in SROs or supportive housing = about 40% to 50% (est., mixed years, definitions differ). The 18% is 246 deaths at City-funded sites in 2020 and 2021 of about 1,368 (press report of Medical Examiner records). The 26% is DPH's 2024 figure, cited by the Grand Jury. As a check, DPH's 2024 split by setting gives 26% + 2% + 1.4% + 19.5% outside = 49%. Sources: OCME 2025 Overdose Report; 2024 PIT Count Report; Controller 2025 Homelessness Needs Assessment; Census county population estimates, 2020-2024; Civil Grand Jury, "At Scale, At Risk" (June 2026); DPH Director's update, Sept 2025, slide 6; SF Standard, April 2022.
              • Housing spending and behavioral health. 53% of single adults in HSH housing programs reported a substance use or mental health disorder (5,647 of 10,592). 53% × $255.1M of site-based housing = about $135M. 53% × $421.6M of all housing = about $223M (est., FY25-26). The 53% is a share of people, not dollars. No source separates drug use from mental health. Sources: Controller 2025 Homelessness Needs Assessment; HSH budget overview (Jan 2026), slide 8.
              • Cost per standard deviation. Cost per student ÷ effect. Tutoring: $1,200 ÷ 0.15 = $8,000; $2,500 ÷ 0.05 = $50,000. General spending: $1,000 in 2018 dollars for four years is about $5,300 today; ÷ 0.0316 = about $165,000; at one-third of that effect, about $500,000. Summer: $1,340 ÷ 0.08 = about $17,000 in 2014 dollars. Saga: $3,500 ÷ 0.40 to $4,300 ÷ 0.18 = $8,800 to $23,900 in 2013-15 dollars (est.). Sources: University of Chicago and MDRC, Personalized Learning Initiative; Kraft, Edwards & Cannata; Jackson and Mackevicius, NBER w28517; NEPC Review of Reber & Goodman; Guryan et al.; Augustine et al.; BLS CPI-U, San Francisco-Oakland-Hayward.
              • Spending and scores. Real spending per unit of attendance rose from about $20,435 (FY18-19) to about $27,642 (FY24-25), about $7,200 in FY25-26 dollars. $7,200 × 285.55 ÷ 363.78 = about $5,650 in 2018 dollars; × 0.0316 SD per $1,000 = about +0.18 SD (est.). In FY21-22, spending rose from $836.0M to $1,017.2M, about 16% after inflation, while attendance fell from 50,227 to 44,794; together, about 30% per student (est.). Sources: CDE Current Expense of Education; BLS CPI-U, San Francisco-Oakland-Hayward; Jackson and Mackevicius, NBER w28517.
              • Cash aid timing. First payment at the earliest: 15 days of verified presence, then an intake, then up to 7 business days. That is roughly 3 to 4 weeks after arrival (est.). Sources: SFHSA CAAP Eligibility Handbook (Oct 2026); SFHSA, Apply for CAAP.
              • Community schools cost. $232,000 a school a year = SFUSD's $33.7M State grant ÷ 29 schools ÷ 5 years (the term is our assumption). $550,000 = the Student Success Fund's top implementation grant. ÷ about 400 students = $580 to $1,375 per student (est.). Sources: SFUSD release on State community schools grants (May 2022); DCYF Student Success Fund grants (March 2024).
              • Lines that are judgments. These (est.) marks are our reading of the data, not a calculation: that part of SFUSD's spending rise came from fewer students being counted and much of it went to benefits, pensions, special education, and contracts; that demographics do not predict higher scores from FY18-19 to FY25-26; that the funnel's outcome is more likely near the bottom of its range; and that the legal outer bound would close shelters, end treatment contracts, and lose Medi-Cal match. The figures behind them are in Why SFUSD's doubled spending hasn't shown up in scores and in Who controls the money.
              • SFUSD check register. SFUSD posts a monthly check register to Board agendas. The PDFs are page images with no text layer, so we ran OCR and rebuilt each row. We matched each month to the Board's cover total: all 12 months of FY25-26 tie, $643,425,635.55 in all. We parsed 17,737 checks against the Board's 17,758. The gap of 21 equals the 21 reissued checks in the registers' summary boxes. The Board's count includes them, and its dollar total does not. Check amounts drive vendor totals, and lines drive fund and object totals. Where OCR lost a line amount or a retention minus sign, we set the line to fit the check amount, which touched $136,775 of lines, under 0.03%. Vendor, capital, self-insurance, and individual payments total $515.1M, within 1% of $517.6M of non-staff spending in SFUSD's accounts, after removing $9.0M in non-cash entries (est.; cash vs. accrual). Sources: SFUSD warrant registers, FY25-26; SFUSD FY25-26 Unaudited Actuals.
              • Who gets paid at SFUSD. The categories are ours (est.). Charter payees by name; capital by fund; self-insurance by fund; benefit plans and deductions by payee and liability code; individuals by name pattern and object 5801; the rest are operating vendors, paid from funds 01, 05, 11, 12, and 13: $301.51M of $643.43M, 46.9%. Top 3 $79.36M, 26.3%; next 7 $54.25M, 18.0%; so the top ten are 44%. Other payees, about 2,000, $167.90M, 55.7%. Service types come from the org name on SFUSD's contract lists, then from object codes. In the list of 367, a row's type follows the largest part of what it was paid. Special education and health providers $66.06M + buses $47.09M + after-school programs $38.54M + school food $21.56M are 57% of operating vendor money. Construction and capital: top 3 $58.40M, next 7 $60.88M, the rest $61.86M. Charter lines: $64.61M from Fund 77, $31.37M coded 01-8096, $3.74M coded 01-7211, $3.54M other. Transportation: Form PCR shows $37.49M of direct cost, all under special education, and $7.20M allocated to regular education by pupils transported, so about 84% goes to special education (est.). About $54.8M of the checks are coded 9510, which appears to mean prior-year bills (est.). The Building Fund's recap total is $162.5M (est.). Payments to banks coded to construction objects look like retention paid to escrow (est.). $3.5M of The Exploratorium's payments are coded to object 01-9512, which we read as a balance sheet account (est.). The type for one row, coded to developer fee revenue, rests on one check coded 25-8681 (est.). Payments to individuals and families from operating funds are $11.0M (est.). Union and benefit plan rows are mostly on liability objects, so we read them as money withheld from pay or owed for staff benefits (est.). Sources: SFUSD warrant registers, FY25-26; SFUSD FY25-26 Unaudited Actuals, Form PCR, PDF pp. 157-158.
              • SFUSD staff share. Salaries and benefits ÷ general fund total, Form 01. FY25-26 actual: $1,005.70M ÷ $1,310.89M = 76.7%. FY24-25 actual: $973.49M ÷ $1,262.75M = 77.1%. FY26-27 budget: $1,094.16M ÷ $1,361.82M = 80.3% (est.). Administrators (objects 1300 and 2300) are $63.12M, 4.8% of the general fund in FY25-26 (est.). The State's CalSTRS payment on SFUSD's behalf, about $41.29M in FY24-25, sits in the CalSTRS line (est.). The move of benefit coding in FY25-26 is our reading of the form (est.). Sources: SFUSD Unaudited Actuals, FY24-25 and FY25-26; SFUSD FY24-25 audit.
              • SFUSD school vs. central figures. General administration $77.85M ÷ $1,310.89M = 5.9% (est.). The FY26-27 central cut of $15.3M at a 15% target implies a base of about $102M (est., rough). Contracts under an SFUSD school's own name are $13.3M, 5.9% of the FY24-25 contract list, at 98 schools. Another $0.8M is under 24 private school names (est.). Sources: SFUSD FY25-26 Unaudited Actuals; SFUSD budget deck, June 2026; SFUSD contract detail list, June 2025.
              • SFUSD all funds. The sum of funds in the FY25-26 Unaudited Actuals and the SFCOE report, $1,742.81M (est.). We leave out self-insurance ($30.13M), because other funds already pay it $35.85M in contributions, and transfers of $137.11M. Indirect charges between funds net to zero. Salaries and benefits $1,090.15M are 62.6% of that total. Without capital, debt, and deferred maintenance, $1,085.63M of $1,428.33M is 76.0% (est.).
              • Private share of K-12 students. Census ACS table B14002, 2020-2024. Share = private ÷ (private + public), kindergarten to grade 12, both sexes. The margin of the share = sqrt(MOE private^2 - share^2 x MOE total^2) ÷ total, the Census method, at the 90% level. San Francisco: 24,959 ÷ 78,981 = 31.6%, with a margin of 1.3 points (est.). The city ranking uses the top 50 places by population age 3 and older, from the same table. Milwaukee is second at 25.2% (+/-1.2). State counts: 24,218 ÷ (24,218 + 56,701 - 3,648) = 31.3%, leaving out Five Keys (est.). Two of every three: 68.4% of SF K-12 residents are in public school, including charters and schools outside SF. State counts, 2024-25: 48,902 district-run ÷ (48,902 + 3,882 charters outside Five Keys + 24,218 private) = 63.5%, or 68.5% counting those charters (est.).
              • Private school enrollment. State affidavit files. Some schools skip a year. If a school files again later, we carried forward its last filing (est.). The 2005-06 to 2024-25 rows are gap-filled, matching schools by State code. As filed: 23,453 in 2019-20 and 24,218 in 2024-25, a rise of 765. We add 1,286 for 7 schools that skipped 2019-20 and 118 for 1 school that skipped 2024-25, giving 24,739 and 24,336, about -400 (est.). None of the 107 State codes in the 2024-25 file is in the State's 2026 test list. District-run schools went from 52,778 to 48,902 (est.). Sources: State private school data; State enrollment.
              • Who attends private school. Census ACS microdata, 2020-2024, K-12 students in SF, with 90% margins from replicate weights (est.). Income is household income in 2024 dollars. Public includes charters and schools outside SF. Private includes home school. Under $100,000: 15.5% (+/-2.6). $500,000 or more: 63.4% (+/-5.2). Race, not counting Hispanic students: 40.6% White and 3.0% Black among private students, 18.4% and 6.5% among public students.
              • UC per 100 12th graders. UC applicants, admits, and enrollees from source schools in SF, Fall 2025, ÷ grade 12 counts from 2024-25: State census files for public schools and affidavits for private schools (est.). Admits: private 1,008 of 2,075; district-run 1,497 of 3,961; district-run without Lowell or Asawa 1,022 of 3,138; charters without Five Keys 113 of 320. UC's file shows no counts under 5 applicants or under 3 admits or enrollees, so totals run slightly low. Of those who applied: private 1,008 of 1,404; district-run 1,497 of 1,838; charters 113 of 145. Sources: UC admissions by source school; State enrollment, 2024-25; State private school data, 2024-25.
              • Charter and district results, pooled. We pooled school rows from the State's research files: students with scores × percent meeting, summed, divided by students with scores (est.). "Charters" means every charter located in SF that year, including closed ones, State Board ones, and Life Learning. Five Keys is separate. County Office schools are left out. Rows under 11 students are suppressed by the State, so they drop out. Low-income means the State's "socioeconomically disadvantaged" group. We use grade rows 3 to 8 and 11, and Student Group 31 for low-income students. Check: pooled district-run results match the State's SFUSD row within 0.1: 53.6 vs 53.50 in reading and 45.9 vs 45.80 in math in 2026. Without Mission Prep, the 2019 charter reading figure falls from 45.9 to 40.7, below the district's 43.4 (est.). Within groups, 2026, all-grades rows: Student Groups 204 (low-income Hispanic), 200 (low-income Black), and 202 (low-income Asian). District-run is the State's SFUSD row. Charters pool school rows, and small rows are suppressed (1 school for Hispanic, 6 for Black). Students with reading scores: Hispanic 5,182 and 1,034; Black 979 and 195; Asian 4,530 and 18. Shares of low-income students tested: 1,034 of 1,480 at charters are Hispanic; at district-run schools, 5,182 of 12,760 are Hispanic and 4,530 are Asian (est.). Source: State test results, SF files, 2019, 2022, and 2026.
              • Charters against similar district schools. All students. Each charter grade group against the pooled result of district-run schools with the same grades and a 2025-26 low-income share within 10 points, leaving out Lowell and Asawa, for groups with 30 or more tested (est.). In 2026: 8 above and 8 below in reading, 7 above and 9 below in math. The count moves with the method: against the median peer school, 10 above and 6 below in reading, 9 above and 7 below in math. Source: State test results, SF file, 2026.
              • Charter enrollment. Ed-Data charts. 2019-20: 8,253 under SFUSD's code + 554 under the State Board - 4,199 Five Keys = 4,608. 2009-10: 2,193 + 468 - 7 = 2,654. District-run = Census Day total - SFUSD-code charters: 55,140 - 2,193 = 52,947; 61,031 - 8,253 = 52,778. Outside Five Keys: (3,901 - 4,608) ÷ 4,608 = -15.3% from 2019-20 (est.). 3,901 ÷ 52,373 = 7.4% of district-run plus charter students (est.). Sources: State enrollment, 2025-26; Ed-Data charter enrollment.
              • One new charter's cost to SFUSD. SFUSD staff's estimate of more than $5 million a year ÷ 396 students = about $12,600 a student (est.). Source: Dragon Gate staff report, Aug 2025, PDF pp. 2, 22.
              • Is 2019 a fair comparison? Mostly yes. The State's cutoff for meeting the standard is the same in its scale score tables for 2019-20 and earlier and for 2020-21 on, in every grade and subject. The 2026 file renames the level "proficient." At the same school mean score, the share meeting the standard differs by 0.3 points or less between 2025 and 2026 (est., California schools with 50 or more tested). The test got shorter in 2020-21, and the State says results can still be compared with 2018-19. Low-income is a yes-or-no State group, and the State applied a parent-education rule to it in fall 2023. It was 54.8% of SFUSD students tested in 2019, 56.6% in 2024, and 54.9% in 2026. Holding race mix at 2019, the 2019 to 2026 change is -3.8 points for low-income students and -0.4 for others (est.). The fall shows in every tested grade. Statewide, the changes are -0.3 and -1.6. Sources: State scale score ranges; State, Understanding Results; CDE 2024 assessment FAQ; CALPADS Flash 260; State test results, SF and statewide files.
              • Low-income students below the reading standard. State test results, 2026, low-income students, all tested grades, at district-run schools. Students below the standard = students tested x (100% - share meeting it), by school, then summed: 7,846 of 12,746 tested (est.). Schools are grouped by 2025-26 low-income share. Under 40%: 30 schools, 2,550 tested, 1,079 below, 57.7% meeting. 40% to 60%: 19 schools, 3,056 tested, 1,599 below, 47.7% meeting. 60% to 75%: 17 schools, 3,116 tested, 2,068 below, 33.6% meeting. Over 75%: 34 schools, 4,024 tested, 3,100 below, 23.0% meeting. That band has 35 open schools; Lee Newcomer (15 students) tested no one in reading in 2026, so it has no reading score. (1,079 + 1,599 + 2,068) / 7,846 = 60%. Two small programs the State holds back, with 14 low-income students tested, aren't counted. The district row shows 7,860 below (12,760 tested, 4,900 meeting). Another 884 low-income students enrolled in these grades have no score. With the same race mix in each group, using State rows that cover 87% of these students, 57.7% and 23.0% become about 52.5% and 30.6% (est.). Sources: State test results, SF file, 2026; State census day, 2025-26.
              • Tutoring all 7,846. 7,846 x $1,200 = about $9.4M. 7,846 x $2,500 = about $19.6M a year (est.). This is the per-student range used in What money can do. It counts tested grades only. Tutoring is about $9.9M: Part 3's $10.0M less about $0.1M for attendance outreach. $9.9M / $2,500 = 3,960 students, about 4,000. At $1,200, $9.9M covers all 7,846, with about $0.5M left (est.).
              • Reading tutoring in 2025-26. DCYF's release (Jan 20, 2026): 1,443 students added with Student Success Fund money, more than 2,700 in all at 20 SFUSD schools, with a focus on grades 2 to 4. The SF Education Fund runs the program, and Chapter One provides the tutors. About $500 a student a year is from press reports. We have not confirmed its 2026-27 funding. The trial: 815 kindergartners in a large Southeastern district were randomly offered the program or usual instruction and followed through grade 3 (2021-22 to 2024-25). 41% of those offered got all four years. Sessions averaged 2.5 a week, about 15 minutes in all, plus practice on a tablet. The program cost about $375 to $465 a student a year. The grade 3 result is imprecise, and the paper is not yet peer reviewed. Evidence for ESSA rates the program "strong." Sources: DCYF, Jan 20, 2026; The 74, Feb 2026 (press); Jung, Robinson, Ripma, Cortes, Kortecamp and Loeb, EdWorkingPapers 26-1548, Aug 2026; Evidence for ESSA, Chapter One.
              • Absence and reading. 2026 reading, 2025-26 chronic absence, and 2025-26 low-income share, by school. Median reading at the 34 schools over 75% low-income: 44.1% where absence is under 30% (6 schools), 19.4% at 30% to 50% (16 schools), 17.4% at 50% or more (12 schools). Median math: 41.6%, 13.4%, 9.4%. Across 100 district-run schools with all three figures, a straight-line fit of reading on low-income share has an R-squared of 0.655. At schools over 75% low-income, median reading is 44.1% where chronic absence is under 30% (6 schools), 19.4% at 30% to 50% (16), and 17.4% at 50% or more (12). Four of those six low-absence schools are 40% or more Asian. Adding chronic absence gives 0.796 (est.). Adding the shares of Asian, Hispanic, Black, English learner, and disability students instead gives 0.876, and adding absence to those gives 0.897 (est.). Of the 34 schools, the 30 under 40% Asian have medians of 20.1% (2 schools), 19.4% (16), and 17.4% (12). Without 4 alternative and continuation schools, the medians are 44.1%, 20.2%, and 18.0%. With 2024-25 absence (6, 19, and 9 schools), they are 44.1%, 18.7%, and 16.8%. That is a statistical fit, not a cause. Sources: State test results, 2026; State chronic absence, 2025-26 and 2024-25; State census day, 2025-26.
              • One percent more attendance. 1% of 45,242 funded units of attendance x $14,812 of State formula money per unit (both FY26-27, SFUSD's budget, PDF p. 5) = about $6.7M a year at full effect (est.). FY24-25 attendance of 45,086 gives the same rounded figure. The $14,812 is an average across the whole formula, so the gain from one more unit could differ. One percentage point on the attendance rate is about 483 units (1% of 48,267 students), about $7.1M. SFUSD's funding uses a three-year attendance average, so the money follows slowly. Sources: SFUSD 2026-27 Recommended Budget, PDF p. 5; State Current Expense of Education, FY24-25; Education Code 42238.05.
              • QTEA a year. $110.1M of parcel taxes in SFUSD's FY26-27 budget ($110,087,486, PDF p. 22) - $57.3M Fair Wages for Educators tax = about $53M a year (est.). SFUSD's June 2024 release says this line holds the two taxes. For a direct check, SFUSD's FY23-24 budget projected QTEA at $49.6M and Fair Wages at $51.5M (Volume II, PDF p. 22), which add up to that year's $101.1M. The Controller's "two other parcel taxes" appear to include a Mello-Roos tax in a capital fund ($8.2M in FY25-26, Unaudited Actuals, Form 49, PDF p. 100), not in this line. The $110.1M is a budget figure; FY25-26 actual was $104.9M, or about $51M after $54.0M of Fair Wages. Sources: SFUSD 2026-27 Recommended Budget; Controller Revenue Letter, June 2026, p. 38; SFUSD, June 2024; SFUSD FY23-24 budget, Volume II.
              • Student Success Fund, year by year. FY23-24: $11.0M appropriated, $9,008,250 agreed, $1,789,131.88 paid (contract 1000031492). FY24-25: $35.0M appropriated, $26,519,360 agreed (contract 1000032928), $34,797,177 allocated and $21,665,719 spent per the annual report, $22,084,862.10 paid and $2,211,622.31 pending per City vouchers. FY25-26: $35.0M appropriated, $28,996,871 agreed, $22,506,802.97 paid so far, $6,490,068.03 still set aside (contract 1000036345). The Fund also pays for City administration and contracts: $5.5M of $45.0M in DCYF's FY26-27 request. The Controller reports reserve deposits of $1.5M in FY23-24 and $16.6M in FY24-25. We could not tie each year's deposit to these figures. The report's District Innovation balance ($4,971,062) looks like a misprint: $7,250,000 less $2,825,539 is $4,424,461. Sources: City contract data; City voucher data; Student Success Fund 2024-25 Annual Report, PDF pp. 32 and 34; Controller Revenue Letter, June 2026, p. 31; DCYF budget submission, PDF p. 45.
              • Staffing by school low-income share. Enrollment / teacher full-time equivalents, and enrollment / pupil services full-time equivalents, summed within each group of district-run schools and four County Office schools, 2025-26 (est.). Under 40% low-income (31 schools): 20.8 and 199. 40% to 60% (20 schools): 19.3 and 161. 60% to 75% (17 schools): 17.4 and 114. Over 75% (38 schools): 17.5 and 82. "Support staff" is the State's pupil services category as Ed-Data reports it: credentialed counselors, nurses, psychologists, social workers, librarians, speech specialists, and other medical staff, shown as whole numbers, so ratios are rough. Classified staff such as paraeducators aren't in it. These roles aren't broken out. Without alternative and continuation schools too: 21.0 and 199 against 18.5 and 84. Elementary schools only: 21.8 and 171 against 20.6 and 87. Sources: Ed-Data, charts 180 and 187; State census day, 2025-26.
              • Empty seats by school low-income share. SFUSD's 2022-23 program capacity - 2025-26 enrollment, summed within each group, so a school over capacity offsets one under it (est.). Under 40%: 561. 40% to 60%: 507. 60% to 75%: 1,309. Over 75%: 3,210. Total: 5,587. Schools under 70% of capacity: 0, 2, 0, and 12, or 14 in all. Capacity and enrollment are from different years. Ruth Asawa and The Academy share one capacity figure on the McAteer campus (1,065 seats, 762 enrolled) and aren't counted, and County Office schools have no capacity figure. Sources: SFUSD 2023 Facilities Master Plan, PDF pp. 76-81; State census day, 2025-26.
              • Gaps. No source totals City substance use costs across all departments. SF's own opioid settlement use clauses, current settlement balances, and SFUSD's Medi-Cal school billing revenue are not public.

              What we couldn't verify

              We rechecked figures against sources in November 2026. These are the items we could not confirm, or could confirm only in part.

              • Final budget line items. We checked the final Budget and Appropriation Ordinance (File 260596, Ordinance 140-26). HSH General Fund, DPH behavioral health, Jail Health, and the FY27-28 fund balance changed slightly from the proposal, and this page uses the final figures. FY26-27 PEEF, the Student Success Fund, and $465.2M of Prop C revenue did not change. The $374.6M of Prop C-eligible spending is not in the ordinance and remains the Mayor's proposal.
              • PEEF in the Budget Book. The Budget Book's $134M is DCYF's whole Public Education fund: $116.3M PEEF, $16.5M PEEF baseline, and $1.45M soda tax. The PEEF line passes $107.6M because it also spends unspent prior-year balances.
              • The reserve. The $18.1M is the balance at June 30, 2025, the last one the Controller reported. The June 30, 2026 balance is not published. City data suggest it could be anywhere from $18.1M to about $28.5M (our estimate, under the Charter's $40M cap), depending on how the Controller treats money from a freeze year. The top end adds up to about $6.0M not granted in FY25-26 ($35.0M appropriated, grant capped at $28,996,871), though part of that can pay City overhead and contracts, and about $4.4M unpaid on the FY24-25 grant ($26,519,360 agreed, $22,084,862 paid, contract 1000032928), though $2.2M of that is listed as pending and some may already be in the $16.6M deposit. Sources: Controller Nine-Month Report, FY25-26; Board Resolution 471-25; Controller Supplier Contracts, DataSF cqi5-hm2d.
              • Student Success grants. We did not find a FY26-27 grant to SFUSD. DCYF's February request budgets about $39.5M for grants.
              • The deficit. We found nothing newer than the Controller's June 2026 letter on FY28-29. The FY27-28 threshold for a freeze is not published yet. The December 2026 plan is not out.
              • Our dollar ranges. The realistic ranges in Who controls the money are our judgment, and no document can confirm them. Jail Health's General Fund share uses DPH's February request, because the ordinance has no fund split by division. The Fire teams' $18.0M sums three FY25-26 project budgets from a February 2024 book; a FY26-27 total is not published. Resolution 63-26 sets the RESET ceiling at $14,537,426 from February 1, 2026 through March 31, 2028. An earlier line that DPH had "already cut about $20M" could not be traced to one source, so we removed it. The prosecution share in $215M is the Budget and Legislative Analyst's 2019 midpoint, which we did not update. Only the preliminary HHAP Round 7 list was found.
              • Housing and bed counts. HSH's vacancy dashboard was rechecked on Nov 1, 2026 and matches 771, 9,160, 361, 332, 78, and the 8.1% to 9.1% range. HSH's 7% goal is confirmed, but we could not check every month; the Budget and Legislative Analyst reported 7.1% on January 31, 2024. DPH's bed finder counts for October 30-31, 2026 come from a live page with no archive.
              • Overdose counts. 2020 to 2025 are Medical Examiner annual reports. 2017 to 2019 are DPH counts from death records, which run within 2% of the Medical Examiner in 2020 to 2024.
              • DPH vendor totals. The behavioral health vendor rows match City data, but we did not recheck how every smaller vendor was classified in the $392M and $405M totals.
              • Press we could not fully read. Add-backs of $41M and about $28M rest on press and a Supervisor's column. The $109 cash grant of January 2025 rests on press and a City page.
              • Housing First as policy. The Budget and Legislative Analyst (May 2024, PDF p. 20) says the City adopted Housing First in its 2004 plan. We did not open the 2004 plan itself.
              • Studies. RAND's NYC community schools report loaded on November 1, 2026 and gives percentage points (PDF pp. 54 and 81). $580 to $1,375 is our own estimate. For Rogers and Feller, the supplement confirms $6.60 per household (Table 29) and 1.1 fewer absences for the two best letters (Table S13). About $6 a day is our division. Evans's 88% comes from secondary summaries; the Science text was blocked. The published text of Kraft, Schueler and Falken was blocked, so the details for programs of 1,000 or more students come from the October 2024 working paper. We read only abstracts of Nickow et al. and Robinson et al. The Personalized Learning Initiative's latest report is June 2025 and preliminary. D.C.'s full tutoring results are not out. We found no independent trial of attendance mailers since the pandemic, only vendor slides, and no state tutoring evaluations beyond those listed.
              • Dollar years. Jackson and Mackevicius's $1,000 is in 2018 dollars, about $1,300 today, and the $50M comparison now adjusts for this. That base comes from their working paper. The Personalized Learning Initiative's tutoring prices are in 2023 dollars, about 11% below today's. Saga's are 2013-15 dollars.
              • School figures. Homeless students' scores: Ed-Data shows 25.81% and 19.07%, and SFUSD shows 26% and 19%. The State comparison for 2016 to 2018 and 2022 to 2024 rests on SFUSD's statement. We could not find how the State treated the February 2026 strike days in chronic absence, which also affects the 13,700. Stanford's business school reported in April 2024 that EveryDay Labs, which sells a mailer program, counted San Francisco among its districts. We could not confirm this with SFUSD. DCYF's 2023 Beacon evaluation would not load. The SFUSD 2026-27 budget was re-fetched live on Nov 1, 2026 and matches the saved copy. Eight other SFUSD board documents were re-fetched on November 1, 2026 and also match: the Facilities Master Plan, the June 2024 plan and Exhibit A, the Central Office plan, the Layoffs Overview, the June 2025 deck, and the 2nd and 3rd Interim decks.
              • Student survey. SFUSD's fall 2024 high school deck shows the share who felt sad or hopeless falling from 35.3% to 24.3%, but its trend note says there was no change from 2021 to 2024. We report the numbers and found no significance test for that drop. SFUSD gives no school-level results.
              • Unit costs. Teacher $161,000 starts from Ed-Data's FY24-25 average salary and our estimate of its growth. Paraeducator $58,000 to $87,000 fits the UESF schedule, but no document gives SFUSD's paraeducator hours or days. The 99 school social workers rest on the SF Standard, June 2026. EdSurge's $1,200 to $2,500 range links to Kraft and Falken (2021). We searched the ERIC copy and did not find the range. That paper estimates $662 to $1,486 by grade range for its own model, which uses high school and college students as tutors, and $2,500 to $3,800 for earlier models with two students per tutor. What money can do cites evaluations at about $1,200 to $2,000.
              • Law and State rules. We read the codified Charter, Business and Tax Code Article 28, DHCS's CYBHI and block grant documents, and current Welfare and Institutions Code sections 5840, 5891, 5892, and 5963.02. A Board file search found no ordinance moving the Sec. 16.133 date. A 2025 program note says an SFUSD and DCYF data-sharing agreement runs through June 30, 2027; whether it meets Sec. 16.133 needs City Attorney review. California's methadone bulletin would not load; we relied on the federal rule. We read Education Code 42238.05 on a public mirror. The timing of the annual behavioral health update is in DHCS's BHSA County Policy Manual, Ch. 3. The City Attorney's 2025 Prop C memo is known only through press. We read the 2002 Prop N text and found no amendment clause, but later ordinances amended that section, and we can't tell which added the rule.
              • Questions for a lawyer. Does a freeze move the Charter's schedule back a year? What base does the 3% cap start from? Does a reserve draw need an actual freeze or cut? Where do unspent draws go? Can the City budget a FY28-29 freeze before the March 2028 test? Is Prop C's base a floor on the General Fund budget? Does Sec. 16.133 reach Parts 1, 2, and 5, and do grants to a school district need gift-clause findings? Do SF's own settlement clauses allow school programs? What vote does a Board-placed change of Prop C uses need, and would State Prop 43 reach it? Can Care Not Cash's baseline change without voters? Can a District Innovation Grant under Charter Sec. 16.131(e)(4) pay for a $10M tutoring program across many schools, and must every school it covers have a Community School Coordinator?
              • Settlement and prevention money. SF's own opioid settlement use clauses and balances are not public; the ordinances and packets we opened hold no use clause, and the budget shows one line, so we can't say which settlements fund it. SF's block grant prevention amount was not found. DPH's decks do not say which four programs make up the $973K, or why each ends.
              • Other secondary items. HSH's survey dashboard states the base of the 2024 count's 32%: 414 people housed in SF before becoming homeless. For 2022 the dashboard (13%, 46%) and the 2022 report (17%, 35%) differ.
              • Data sites. The 31 site figures marked unverified, and the live 911 data, were not re-pulled.
              • Contacts. We did not test delivery to the Supervisors' First.Last email addresses.
              • Our own analysis. DPH's Health Commission reports show a 12% contingency on a methadone contract under the same RFP as BAART; we did not find BAART's own figure. That ceilings are set high for Medi-Cal-billed volume is our analysis with no source document.
              • SFUSD payments. We could not find vendor payments before July 2025, except a short 2023 register; FY23-24 and FY24-25 have contract lists only. We found no FY25-26 year-to-date contract list. There is no total lifetime cost for EMPower, because approvals overlap and payments before July 2025 are not public by vendor. We could not find the purpose of The Exploratorium payments, SFUSD's list of local sub-object codes, pension and health plan remittances by payee, or total spending by school with central services allocated. FCMAT and State adviser reports do not report vendor-level data.
              • Private and charter schools. We could not find school-level test results for any SF private school, NAEP results for private schools in SF or California, or low-income, English learner, or disability shares by private school. We could not find an average SF private tuition or total aid across all schools, SF residents enrolled in charters outside SF, SF private students living elsewhere, or Five Keys students by county. We could not find SFUSD's total oversight fees, Prop 39 offers, pro rata rate, or total parcel tax paid to charters, or The New School's current renewal status. The State's private school FAQ and the digests for Prop G and Prop J would not open.
              • Dates. The City's 2027 calendar is not published, so most 2027 dates follow the prior cycle.