From this project’s research library · receipts on every claim

The Core Facts: Homelessness in Toronto

Updated August 2026 · Every figure on this page links to its primary source. The deeper receipt chains live in our research library — each section names its backgrounder. Built for anyone making the public case: deputations, letters to councillors, newsletters, conversations. Use the exact figures, keep the years attached, and never mix the denominators (§4 explains why).

The whole page in one paragraph: Toronto spends about $900 million a year on homelessness — $41,000 to $100,000 per homeless person depending on how you count — to rent people a bed for the night at up to $491 a night, under contracts with no requirement that anyone ever exits to housing, at prices the City stopped publishing in 2012. Housing a person costs $6,311 a year net. The math doesn’t work — and the City’s own documents are the proof.

On this page

• 1 · How many people (the number ladder)
• 2 · What we spend
• 3 · What a night costs
• 4 · “$900M for 12,000 people” — the real math
• 5 · What the money doesn’t buy
• 6 · The deaths
• 7 · What we found when we pulled the thread
• 8 · Who it lands on
• 9 · The fix costs less than the failure
• 10 · What nobody will tell you — the gaps
• 11 · Claims we checked and won’t use
• The receipt shelf — download every source

1 · How many people — the number ladder

Toronto has no single homelessness number. It has a ladder, and every rung is real. Using the wrong rung is how advocates get dismissed — so here is the whole ladder:

• 12,196 — people experiencing homelessness on the night of October 22, 2025, the City’s first stand-alone point-in-time count.
• 15,418 — the October 2024 Street Needs Assessment, more than double the 7,300 of April 2021. Breakdown: 9,026 in base shelters, 1,446 outdoors, 1,724 in provincial settings (corrections, health, violence-against-women shelters).
• The 21% one-year “drop” is a refugee-policy story, not a success story. It tracks the refugee-claimant caseload ebbing (~6,600 to ~3,500 nightly, Aug 2024 to Aug 2025). Chronic homelessness did not improve — and the share of people sleeping outdoors roughly doubled in Etobicoke-York (7%→14%) and North York (5%→10%).
• The City’s own pages disagree with each other: “actively homeless” has appeared as both 10,569 and 9,251 for the same metric and month (May 2026); public health’s flow-data figure was 11,094 (January 2026). Unresolved, on the City’s own surfaces.
• Every official count excludes hidden homelessness by design. The City’s own methodology page says the count “did not include people experiencing hidden homelessness, such as people temporarily staying with friends or family.” The City’s By-Name List separately discloses an 18–20% structural undercount.
• A reasonable all-in estimate is 35,000–45,000 people (central range 38,000–42,000). Two independent methods converge there: Ontario’s own leaked internal estimate of 234,000 homeless provincewide implies ~40,700 for Toronto at per-capita rate, and standard hidden-homelessness multipliers (2.0×–2.9×) on the official count give ~30,800–44,700. The province called its own 234,000 figure “inaccurate” — and never offered a corrected one. Two Ontario-government-derived estimates for the same year disagree by nine times.
• Ontario-wide: 84,973 known homeless (2025) — 53% chronic, and chronic homelessness has roughly doubled since 2019.
• Sleeping outside is a nightly reality in the thousands: 1,446 people were counted outdoors on one October night — a floor, not a ceiling, by the count’s own design. And there is documented precedent for a single downtown site: through summer 2023, up to ~100 asylum seekers, predominantly Black and African, slept on the sidewalk outside the City’s own intake office, night after night, as the direct result of a City shelter-access decision. No official dataset counts who sleeps at City Hall or any civic square on a given night — that gap is one of our standing asks.

Receipts: City of Toronto, 2024–2025 SNA Key Highlights (PDF) · 2024 Street Needs Assessment full report (PDF) · City SNA methodology page · The Trillium reporting · AMO/HelpSeeker, Municipalities Under Pressure (2026 update, PDF) · Ombudsman Toronto investigation. Deep dive: our backgrounder on how Toronto counts — and undercounts.

2 · What we spend

• Toronto Shelter and Support Services, 2025: $897.957 million gross ($656.778M in federal/provincial/other revenue, $241.178M net City property tax; 1,497.5 staff). 2026 budget: $786.068M gross. 2024: $846.313M budgeted, $796.40M actual.
• Who pays: federal $311.2M (34.7%), provincial $286.1M (31.9%), City tax $241.2M (26.9%) — and $59.5M (6.6%) the City’s own budget note never labels. The published revenue breakdown does not sum to the published gross. That unlabeled remainder is itself a finding.
• More than a third of 2025 was refugee response: $321.7M. The federal reimbursement then fell off a cliff: $300M (2025) → $97.1M (2026) — minus 68%, a $203M hole. Ottawa offered 26% of Toronto’s 2025 funding request while other municipalities collectively received about 90% of theirs.
• The housing-subsidy lever was cut while shelter spending grew: the Canada-Ontario Housing Benefit for Toronto went $38M → $19.75M → $7.95M in two budget cycles (−79%). During its April–October 2024 suspension, housing exits from shelter fell 26.7%.
• The City builds shelters almost alone: its 10-year shelter capital plan is $957.1M — 72% City debt-financed; federal and provincial contributions total 1.2%. The new-shelter program prices out at $421,875 per bed in capital cost ( $674.5M for 1,600 beds), 13 of 20 sites funded, first opening 2027.
• Beyond the shelter budget the meter keeps running: excess healthcare costs of Toronto homelessness are estimated at $69.8M–$99.7M a year (a homeless person’s public healthcare costs average $12,209/yr vs $1,769 for a housed person — 6.9 times higher). Private security contracts tied to shelters and encampments: $109M contracted ceiling, $35.2M spent in 2025 alone. A single 2021 clearing of three park encampments cost about $2M. All sources combined, Toronto’s spend is in the range of $900 million to $1 billion or more a year.
• Ontario doubled the money and the problem got worse: provincial + municipal housing/homelessness funding went from $2.1B (2018) to over $4.0B (2025). From 2021–2025, funding grew 32% while homelessness grew 49% — and within that funding, emergency-shelter spending rose 51.6% while community-housing spending FELL 0.6%. The system is buying more nights, not more homes.

Receipts: TSSS 2025 Budget Notes (PDF) · TSSS 2026 Budget Notes (PDF) · TSSS 2024 Budget Notes (PDF) · Mayor’s letter on the housing-benefit cuts (Council item 2025.EX26.13) · The Cost of Inaction (ICES / Homeless Hub, 2024) · AMO/HelpSeeker 2026 (PDF). Deep dives: full-cost accounting · the federal funding architecture · provincial funding.

3 · What a night costs

• In 2010 the system-wide average was $52.13 a night — published, per shelter, by name. To 2023’s $136 audited base rate: +161% nominal, roughly +96% after inflation — over the same years the per-shelter price list stopped being published entirely (§7).
• Comparators: Calgary’s largest shelter operates at roughly $70 per person-night on its own filings; Ottawa reports about $100 per household-night. In the same administrative dataset, Toronto’s median shelter stay ran ~45 days against Calgary’s ~17; chronic-user episodes averaged 329.7 days in Toronto vs 116 in Calgary.
• The alternatives, priced by the City’s own auditor: emergency shelter costs 3× supportive housing and 7–10× a rent subsidy (2022 audits). Housing First support, Toronto-priced in peer-reviewed research: $6,311 per person per year net of the service costs it avoids — against $49,640 for a shelter bed. That is a $43,329-a-year gap, per person. For scale: a provincial jail bed runs $326/day, and a homelessness-coded hospitalization averages $16,785 per stay — 2.15× the national average, with 93% arriving through the emergency department.

Receipts: AG Warming Centres & Winter Respite audit, Feb 2025 (attachment PDF, revised Feb 6 version) · AG Emergency Shelters Part 2 (hotels), 2022 · the 2010 per-diem staff report — the last era of published prices (PDF) · Latimer et al., Psychiatric Services, 2020 · CIHI/ICES hospitalization figures via Homeless Hub. Deep dive: what a bed-night costs — the full escalation history.

4 · “$900 million for 12,000 people — the math doesn’t math.” Here is the actual math.

The instinct is right. The arithmetic needs armour, because the first person you say it to will pick your denominator apart. Divide the 2025 shelter budget by any of the real population figures and you get a monstrous number — but each division answers a different question:

The honest reconciliation, in five steps — this is the version that survives a hostile audience:

1. The population figure is a snapshot; the money serves a flow. Roughly 22,000 unique people move through the system in a year, not 12,196. That correction alone halves the naive figure — and the result is still about $41,000 per person served, every year.
2. A third of the 2025 money was the refugee-response surge ($321.7M), now collapsing along with its federal funding. Strip it out entirely and the core system still ran roughly $576M — the per-person math barely improves.
3. The money buys nights, not exits. It goes to emergency accommodation at hotel prices ($253/night) and warming-centre prices ($336–$491/night), plus $35M+ a year in private security — under contracts in which no shelter operator anywhere in the system is required to meet a housing-exit target. The City pays per bed, per night. Nobody is paid for anyone getting out.
4. So throughput falls as cost rises. People served per bed per year has declined every single year since tracking began in 2011, to 2.02. The cost per bed-night has nearly tripled off its 2010 baseline. In 2023, 10,183 people entered homelessness and 5,927 exited to housing. The machine runs harder, moves fewer people, and charges more per night.
5. And you cannot audit where it lands, because the City stopped publishing the prices. Per-shelter rates — published every year through 2012, when the system cost $60.7M — have not been published since, at $898M. The division the public most wants to do (this operator, this building, this rate, this outcome) is impossible with public data. That impossibility is itself the strongest fact on this page.

Every figure in this section is receipted in §§1–3 above; the per-person divisions are arithmetic on those receipted figures (2023 denominators shown against the 2025 budget where marked — the City publishes no single-year aligned pair, which is part of the point). Deep dives: the economics · cost per bed-night · full-cost accounting.

5 · What the money doesn’t buy

• People turned away nightly, on a worsening curve: average unmatched callers to Central Intake per night across three winters: 43 → 112 → 174 — up 305% in two years; 216 a day by November 2024, in the City’s own federal funding application; still ~50 a night in March 2026. In winter 2023-24, 12,742 shelter bed-nights sat empty, lost to process rules, while an average of 174 people a night went unmatched.
• The real turn-away number is unknown — by the auditor’s own admission. The division does not track walk-in turnaways at all. Staff at multiple winter sites told the Auditor General they turn people away every day, uncounted.
• The 2020-21 freedom-of-information floor: 117 callers a day told no space; 38 people a night still without a bed at 4 a.m.; 3,646 calls disconnected before being answered. Filed as sworn evidence in court.
• The waitlist is a second city: 105,115 households on Toronto’s social-housing waitlist at the end of 2025 (81,664 in 2020). In 2025, 21,538 new applications arrived and 2,838 households were housed — 7.6 applications for every one household housed. Toronto Community Housing’s own affordable-housing waitlist has been closed to new applications since June 2019. Another 14,000+ people wait for supportive housing.
• Cycling, not housing: the Auditor General found 1,340 shelter clients had been admitted to ten or more different shelter programs. Nightly occupancy runs 96–100%; on a typical August 2026 night, 58 of 91 bed-based programs sat at exactly 100%.
• Violence is up and under-reported: physical assaults on shelter staff rose 423.5% from 2011 to 2021. A press-vs-FOI comparison found 2,408 disclosed violent incidents in 2019 against 157 reported in the media — a fifteen-fold gap.
• Outreach without outcomes: transit outreach site visits rose 17% (2023→2024) while successful shelter referrals from those visits fell 76% (517 → 122).

Receipts: AG Feb 2025 audit · AG Emergency Shelters Part 1 (case management) · City Social Housing Waiting List Reports (quarterly) · CAMH Shelter Safety Study, 2024 (PDF) · City open data: nightly occupancy. Deep dives: the Central Intake failure · the waitlist data · capacity strain.

6 · The deaths — the numbers the City had to double

• Toronto Public Health revised its own counts of homeless deaths upward 75–100% after finally integrating Coroner’s data (February 2025): 2022 went from 189 to 331; 2023 from 150 to 300. Treat every recent year’s count as a floor.
• 2024: 215 deaths. Median age at death: 50 for men, 38 for women — against 78 and 85 in the housed population. A homeless woman in Toronto dies, on median, 47 years younger than a housed one. About 55% of deaths are drug toxicity; overdose accounts for 76% of women’s deaths.
• Among shelter residents specifically: 825 deaths from 2007 through 2024 — 59 in 2024, 59 again in 2025. Priced by the federal government’s own mandatory value-of-statistical-life methodology ($9.8M), a single year’s 59 deaths represent a $578M loss — comparable to the entire shelter budget. (Always framed as the government’s own required methodology — never a literal price on a life.)
• Overdoses in shelters roughly doubled year over year: 47 per month (2025) to 96 per month (2026 to date) — and this dataset excludes everyone outdoors, by design. Shelter overdose incidence has been computed at roughly 209 times the general population’s.
• Encampment fire deaths: 7 since 2010, with 216 encampment fires in 2020 alone — and the public updates simply stop after 2020. No one currently publishes how many people have died in Toronto encampment fires since.

Receipts: TPH, Health Impacts of Homelessness — Board of Health report, March 2026 (PDF) · City open data: Deaths of Shelter Residents · City open data: Suspected Opioid Overdoses in Shelters · Ali et al., BMC Public Health, 2026 (DOI 10.1186/s12889-026-27460-8). Deep dive: the mortality and overdose record.

7 · What we found when we pulled the thread

The transparency collapse

• Until 2012, the City published a shelter-by-shelter price list — every operator, every rate ($20.25 to $75.75 a night in 2010). When the provincial funding model changed in 2013, five figures vanished from public view and never returned: the per-shelter rate, the system-average rate, the Province’s share, the projected shortfall, and per-shelter occupancy. The system then grew from $60.7M to $898M with no published unit prices. Council never voted to stop publishing; the numbers just stopped appearing.
• The 2016 funding model was adopted as explicitly “interim… until such time that City Council approves a final funding model.” No final model has ever been confirmed as brought to Council. A decade of purchase-of-service contracts has run on an interim basis.
• Single-shelter proof: for one long-running 60-bed downtown shelter, the City published the funding figure every year from 2011 to 2020 — to the cent — then nothing for 2021–2026. Today’s cost can only be estimated (roughly $1.9M–$2.6M a year). The City knows the actual number and does not publish it. That is the finding — not the estimate.
• A 250–300-bed winter respite program ran three consecutive seasons at Exhibition Place without a single published dollar figure — and without any Council item authorizing it (it ran under delegated staff authority). The Auditor General’s audit of that program class found the City paid winter operators the full $12.5M in March 2024 against $9.7M in reported costs, with ~$2.9M in surpluses still unrecovered seven months later. The audit names no operator.
• The City’s own 2025 all-of-city homelessness program inventory — 330 programs across 11 divisions — contains not one dollar figure. No budget line, no cost-per-client, no summable capacity, anywhere in the document.
• Of those 330 programs, exactly five are “Employment & Training” — and three of them are recreation youth-referral schemes; one trains shelter staff, not residents.

The auditor’s findings

• $13.2M (about $15M with tax) overpaid to hotels in two years for charges the contracts did not permit — including $5.4M paying full nightly rates on vacant rooms. The auditor’s own comparison: enough to house 140 people with full supports for a year — roughly 52,000 room-nights.
• Modular housing: 63% ($33M) over budget, 159 of 275 promised units delivered — and the Auditor General was “unable to conclude” it was faster or cheaper than conventional building.
• No standardized oversight of $42M+ in shelter security contracts; guards paid about $18/hour and billed out at $25–$60/hour. One security staffing contractor holding $19.5M+ in City contracts faces a proposed wage-theft class action alleging years of unpaid overtime — an allegation before the courts, not an adjudicated finding.
• Ten non-competitive blanket contracts against two competitive ones for hotel-shelter services; single-source hotel catering deals at $20M and $25M; three custodial contracts worth $50M combined surfaced only when a “confidential” document turned out to be mislabeled and public.

The governance record

• Toronto’s Ombudsman found the City’s 2022 exclusion of refugee claimants from shelter beds amounted to systemic discrimination — “specifically, anti-Black racism.” The City Manager formally disagreed in writing. Council initially received the 122-page report without adopting any of its 14 recommendations, and voted 14–9 against reconsidering. Through that summer, up to ~100 asylum seekers slept outside the intake office. A proposed $50M class action followed and is live. Implementation finally began under a 2025 Council item.
• December 2017: Council voted 17–25 against asking for the armouries as winter shelter — with the federal government already agreeing in writing — in a year with nearly 100 homeless deaths and 95% shelter occupancy. Moss Park armoury opened months later after public outcry anyway.
• 2013: a staff recommendation for an Ombudsman investigation of the entire shelter system was deleted on the Council floor (“these activities are currently underway by staff”). Independent oversight took another decade to arrive.
• A 2017 Ombudsman recommendation on over-capacity protocols was still unimplemented at the February 2025 audit — eight years later.
• The oversight blind spot is structural: two of the system’s largest shelter operators appear nowhere on Ontario’s public-sector salary disclosure — the statute keys to provincial funding and they are overwhelmingly City-funded. One major youth-shelter operator holds $48.5M in reserves — 1.55 years of its spending — while running at 98% occupancy and turning young people away.
• The courts have started saying it out loud: a 2026 Ontario Superior Court ruling (Region of Waterloo, 2026 ONSC 2971) held that homelessness itself can be a protected ground under the Charter’s equality section — built on a documented gap of 2,371 unhoused people against 377 shelter spaces. It is under appeal; the Premier called it “the most ridiculous” ruling he had ever seen.

Receipts: the 2010 per-diem report — the last published price list (PDF) and the allocation series that replaced it (2011 · 2012 · 2013 · 2014 · 2015 · 2016 · 2017 · 2018) · AG hotels audit, 2022 · AG winter programs audit, Feb 2025 · Ombudsman Toronto investigation. Deep dives: accountability & complaints · procurement & subcontracting · security contracting · operator financial profiles · legal rights frameworks.

8 · Who it lands on

• Indigenous people: 3 to 16 times overrepresented in every count with an Indigenous breakdown — 9% of the 2024 count against ~3% of the population, 31% of people sleeping outdoors, and peer-reviewed research finding 27.3% of First Nations, Inuit and Métis adults in Toronto have experienced homelessness.
• Black Torontonians: 51.6% of the 2024 count were refugee claimants, predominantly African; the shelter-exclusion decision drew the City’s own anti-Black-racism finding (§7); one major youth agency reports Black youth are 43% of the young people it serves, against 9.5% of the population.
• Women: 41% of the homeless population, 18% of adult shelter beds — only 23 of 93 adult shelters are women-only, with the lowest harm-reduction availability of any shelter type — and a median age at death of 38.
• Youth: 1,500+ homeless 16-to-24-year-olds on a given night against roughly 740 youth shelter beds — more than two to one. Nationally, 42% of homeless youth report having attempted suicide.
• The jail pipeline: 7,455 people released from Ontario jails with no fixed address in a single year (2023-24) — 80% of those held were on remand, never convicted — and the rate of release into homelessness nearly doubled from 2016 to 2022. This discharge-planning failure was first documented in 2006.
• The income floor is the trapdoor: a single adult on Ontario Works receives $733 a month — and someone in a tent, unable to prove rent costs, receives the $343-a-month basic-needs portion only: 13 cents of every poverty-line dollar. The public purse pays that person $343 a month — and $136 a night ($49,640 a year) for the shelter bed they are pushed into.

Receipts: 2024 SNA (PDF) · John Howard Society of Ontario, From Incarceration to Encampment (2025), via CBC · Ali et al., BMC Public Health, 2026 · Ontario Works Policy Directives 6.1/6.3. Deep dives: Indigenous overrepresentation · the refugee-claimant exclusion · gender & the shelter system · youth shelter capacity · root causes.

9 · The fix costs less than the failure

• Housing First works, and the evidence is Canadian: the largest randomized trial in Canadian history found people housed under Housing First spent 73% of their time stably housed against 32% under usual care; the Toronto high-needs group returned $15.05 for every $10 invested. Scotland, at national scale: 2,154 people housed, 83% still housed.
• Ending Toronto homelessness is a capital project the City already outspends: our rebuilt bottom-up costing (it replaced a circulating figure that failed our own arithmetic audit — see §11) prices housing the entire October-2025 population at $1.0B–$1.5B in one-time capital plus about $242M a year ongoing in the balanced scenario — with a pessimistic ceiling of $2.24B. Every scenario costs less than three years of status-quo spending.
• The steady-state prize: roughly $521M a year in avoided shelter costs at the chronic population’s scale (sensitivity range $220M–$620M).
• Prevention runs 10–25 to one: a first shelter month costs about $4,100; a $2,000 rent-bank grant can prevent the entry entirely. And the Auditor General priced a zero-construction win back in 2019: fixing vacancy management at Toronto Community Housing would house 2,200 more people and recover $7M a year — without building anything.
• The comparators, stated honestly — this is the version that survives hostile review: Finland cut shelter use 76%, then saw homelessness rise 20% in 2025 when rents outran support funding. Houston cut homelessness 63% since 2011 within existing federal money — then turned to mass ticketing. Medicine Hat reached functional zero in 2021 and lost it within 18 months. The pattern in every case is durability-of-funding failure, not model failure. No one has shown Housing First stops working; they have shown it stops being funded.
• The 27-year receipt: since the 1999 Golden Report — the blueprint Toronto commissioned and praised — no combination of the three governments has ever funded the known solution at the scale and simultaneity it requires. The missing input is not knowledge.

Receipts: Stergiopoulos et al., JAMA, 2015 (At Home/Chez Soi) · Latimer et al., 2020 · AG, Opening Doors to Stable Housing, 2019 · the 1999 Golden Report on Council’s own record (PDF). Deep dives: the Housing First evidence · functional-zero case studies · international comparators · the book’s full costing chapter · the master recommendations.

10 · What nobody will tell you — the standing information gaps

Every line below is a documented refusal or absence, not an oversight we suspect. Each one is usable verbatim as a freedom-of-information request, a councillor question, or a deputation line — the asks page carries ready-to-use wording for all twelve.

1. Per-shelter rates, 2013 to present (last published: 2012).
2. Any dollar figure, for any season, for the Exhibition Place winter respite — and the missing Council authorization.
3. Current funding for any named shelter (published to the cent until 2020; silence since).
4. The “final” shelter funding model Council was promised in 2016.
5. Walk-in turnaway counts — the City’s auditor says they are simply not tracked.
6. Encampment fire deaths since 2020 — public tracking stopped.
7. The unlabeled $59.5M in the shelter division’s own 2025 revenue breakdown.
8. A complete list of all ~75–81 contracted shelter operators with contract values.
9. The City’s own 330-program homelessness inventory, with dollars attached.
10. The corrections-to-homelessness overlap: 7,455 no-fixed-address releases a year, never costed.
11. Nightly counts of people sleeping rough at civic sites — City Hall, Nathan Phillips Square, Metro Hall. No such dataset exists.
12. Hidden homelessness — excluded from every official Toronto count by design; the province’s own internal estimate was disowned without a replacement.

11 · Claims we checked and won’t use — and neither should you

We ran every hard-hitting claim we could find through adversarial review against primary sources. These circulate widely and failed. Using them hands your audience the counter-punch:

• “$54k a year for a shelter bed vs $7.5k for Housing First” — no primary source. Use $49,640 vs $6,311 net (§3): both trace to audited or peer-reviewed sources.
• “Housing First costs a third of sheltering” — arithmetically false on the real figures. Use the dollar pair.
• “$2.4–$2.7 billion to end homelessness” — doesn’t multiply out from its own inputs. Use the $1.0B–$1.5B scenario range (§9).
• “Supportive housing costs $613 a month” — traces only to an unsourced social-media graphic.
• “40–50% of Toronto’s homeless are Indigenous” — unsupported. Use the receipted 3–16× overrepresentation figures (§8).
• “Shelters feed people on $6 a day” — not found in any source we could locate.
• “Executives earn six figures while charging $54k a bed” — a petition’s framing whose per-person arithmetic deliberately excludes refugee claimants from the denominator to inflate the result. The real, separately-receipted facts (§7) are strong enough without it.
• “163 women killed by gender-based violence in Toronto in 2024” — no primary source locatable.
• “92,965 people on the housing waitlist” — that figure counts applications/households, not people. Say “105,115 households” (year-end 2025).
• Crediting the 2025 count drop to City management — contradicted by the evidence; it tracks the refugee-claimant caseload (§1).
• “The shelter budget is $912M” — the adopted 2025 figure is $897.957M.
• Mixing the three encampment metrics (sites citywide; sites on City property; people in park tents) — they are different objects moving differently. Cite one, labeled.

Check everything — the receipt shelf

Found an error? We correct in public: send it through the Harvest and the fix propagates to every page that cites the figure — that is our commitment.