How the rankings are built

Sources

• Traffic counts — City of Toronto, Traffic Volumes – Multimodal Intersection Turning Movement Counts, "most recent summary" table: one row per intersection with the latest count's totals by mode and approach. Refreshed daily by the City; this build uses counts through Sep 29, 2026.
• Serious-injury collisions — City of Toronto, Motor Vehicle Collisions involving Killed or Seriously Injured Persons, sourced from Toronto Police Service. One row per involved person since 2006; this build uses records through Oct 11, 2026.
• Red-light cameras — City of Toronto, Red Light Cameras. 301 locations.
• All collisions — Toronto Police Service, Traffic Collisions Open Data: every collision reported to police since 2014 — property damage only, fail-to-remain, injury and fatal — one row per event with its location and who was involved. 841,092 events in this build, 567,060 within 65 m of a counted intersection.

Two ways to count collisions

The most collisions rankings use the police dataset — the big number, dominated by property-damage collisions, which tracks volume closely. The most dangerous rankings use the City's KSI dataset — only collisions in which someone was killed or seriously injured, a far smaller number that says more about consequence than frequency. Every intersection page shows both, and neither is a statement about fault.

Names

The City names each node by every road that meets there ("Spadina Ave / Bloor St W / Spadina Rd"). Pages use the spoken two-street form — "Bloor & Spadina" — chosen by rule: ramps, trails and river names are dropped, a road that continues under a different name is counted once, east/west or north/south halves are merged, the more prominent street (the one appearing at more intersections citywide) goes first, and a short list of well-known pairs overrides that order ("Jane & Finch", "King & Spadina"). The formal name and the City's node name appear under every heading.

What an intersection is

An intersection is a node in the City's centreline network. Its name is the City's own (e.g. "Kennedy Rd / Lawrence Ave E"), shown here with an ampersand. An intersection gets a page when it has a traffic count dated Jan 1, 2015 or later, or at least one KSI collision matched to it. 5,838 of 6,122 counted intersections qualify in this build.

Matching collisions to intersections

Each collision record carries coordinates. KSI person-level rows are first combined into one collision per collision ID; police records are deduplicated by event ID. A collision is attributed to the nearest intersection node within 65 metres — roughly the width of a six-lane arterial crossing — whether the police coded it as intersection-related or mid-block; beyond 65 m it is attributed to no intersection. In this build 5,253 of 7,668 KSI collisions matched an intersection.

A collision counts as fatal if any involved person's record is classed Fatal; as involving a pedestrian / cyclist / motorcyclist if the City flags that road user in the collision; and as in darkness if the light condition is recorded as dark (with or without street lighting).

Windows and comparability

• A rank is shown only while it means something. Each ranking covers just the intersections with a non-zero figure — the KSI ranking, for instance, contains only the corners where someone has been killed or seriously injured, not every intersection in the city. So a position near the bottom of one of those lists is not a good sign; it means the corner is on a list that most corners never join, and hundreds of corners share the same count down there. Positions in the bottom half of a ranking are left off the page for that reason: the count itself already says everything the position could.
• Collision rankings use the last ten complete years, 2016–2025. Records can take a year or more to be finalized, so the current year is never ranked. All-time counts since 2006 appear on each page for context.
• Busiest rankings use 14-hour counts only. Before September 2023 the City counted for 8 non-continuous hours; since then, 14 continuous hours. The two are not comparable, so older counts are shown on intersection pages but not ranked. Each figure is one day's count, not an annual average.
• Highway ramps and collector-lane nodes (names containing "Hwy", "Ramp", "Collector", "Express") are excluded from the busiest rankings because they are not what people mean by a busy intersection. Around 140 records whose names are machine-generated road geometry rather than a place — "Hwy 401 Collectors E Yonge St N Ramp & Lord Seaton W Yonge Ramp & Yonge St" — also carry noindex and are left out of the sitemap. They are still published, still linked and still reachable; they are simply not offered to search engines as pages anyone was looking for.
• Exposure-adjusted rate (KSI per year per 10,000 daily vehicles) is shown only where an intersection has at least five KSI collisions in the window and a 14-hour count, to avoid ranking on noise.

Red-light cameras

Cameras are joined to intersections by the City's intersection ID, falling back to the nearest node within 40 m. Ontario ended automated speed enforcement on 14 November 2025; speed-camera locations are therefore historical and are not shown as active.

Tickets. The City publishes Red Light Camera Annual Charges as a spreadsheet: one row per camera site, one column per year since 2007. It is joined to the camera list by the City's location code (the spreadsheet numbers sites from 2501, the camera list from 6001 — the same sites, offset by 3,500), with a fallback on the two street names. "Charges" are tickets issued, not convictions. The current year is partial — the file states the cut-off date and the pages repeat it — so rankings use the last complete year. Sites the City has decommissioned stay in the spreadsheet and are listed separately with their totals. In this build 301 of 301 cameras matched a ticket record (0 by street name).

Collisions recorded after the window closed (from January 2026 to the latest date in each dataset) are never hidden: they appear on the intersection page as a separate "since January" line in the cards, tables and text, and a fatal one is stated in the first paragraph. They are simply not part of the ranking, which needs complete years.

"What's here" layers

Each layer is a proximity join from the intersection node, computed at build time from a City dataset. On the 79 corners along Steeles Avenue, the city limit, one side of the street is in Vaughan or Markham and outside every dataset here; those pages say so. Distances are the distance you would walk — see “How far is it?” below.

• Bikeways — a Cycling Network segment passing within 30 m; the segment's highest-order infrastructure type is shown. "Protected" means a cycle track or multi-use trail.
• Getting here (within 1 km) — every agency's GTFS schedule feed is read at build time: TTC (subway stations with their lines, streetcar and bus routes with the nearest stop for each), GO Transit (rail stations with their lines, GO bus routes), UP Express, VIA Rail, and the regional agencies that cross into Toronto — MiWay, Durham Region Transit, Brampton Transit and York Region Transit — where their feeds are available. A route counts if any of its stops is within 1 km; the distance shown is to that nearest stop. Because the feeds describe the timetable in force on the build date, a temporary diversion or replacement bus appears exactly as the agency schedules it — a streetcar route listed on a street it does not normally use is usually that. The "at the corner" chips still use 100 m for stops and 250 m for a station. GO and UP Express data is provided with permission of Metrolinx.
• Within walking distance (1.5 km) — schools, parks, community centres and arenas, Toronto Public Library branches and the City's Places of Interest and Attractions list, each with its distance; counts are of everything within the radius, the nearest few are named. The "near schools" rankings keep the stricter 250 m rule.
• Bike Share — stations within 150 m for the chip, and the nearest dock plus the count within 1 km, from the live GBFS station list.
• Parking (within 2 km) — three sources. Every lot: the City's topographic mapping of parking-lot footprints (about 19,000 polygons drawn from aerial photography — surface lots of any ownership, public or private, but not underground or structured garages unless their footprint is visible); each lot's area is measured and spaces are estimated at one per 30 m², the usual planning rule of thumb, so treat those figures as approximate. Lots within 45 m of a Green P car park are flagged Green P; the rest are private, commercial or institutional. Green P: the Toronto Parking Authority's car-park list (a 2019 snapshot — the City retired the feed) with spaces and average daily peak occupancy from the quarterly Parking Occupancy dataset where the car-park number matches. EV charging: City-operated stations with Level 2 and fast-charge port counts; private networks are not included. Bike parking: post-and-rings within 300 m, racks and corrals with capacity within 600 m, indoor bike stations within 2 km. On-street car parking is not included anywhere.
• Signal features — from Traffic Signals Tabular, joined by node ID then signal number: accessible (audible) pedestrian signal, leading pedestrian interval by direction, transit priority, bicycle signal, activation year.
• Pedestrian crossover — a PXO within 30 m.
• Live traffic camera — the City's RESCU camera list (336 cameras, with a still image the City refreshes every few minutes) joined to the nearest node within 110 m. The page shows the City's 320 × 240 image directly, never enlarged beyond twice its size; a small server function reads the City's Last-Modified header so the card can show how old the frame is and whether a reload brought a new one. The four fixed-direction reference stills (199 × 112) are shown at their own size.
• Measured speeds — the City's midblock speed, volume and classification counts (automatic tube counters laid across a road for one to seven days, recording every vehicle). For each intersection the most recent count on each of its own streets within 400 m is chosen — matched by street name, preferring counts that recorded speed — and shown with its date, daily volume, average, 85th- and 95th-percentile speeds. A midblock reading describes the block, not the movement through the intersection, and the City counts on request rather than on a schedule, so some readings are years old. A count that recorded under a quarter of what another count on the same street recorded is treated as a partial count — one lane or direction, or a counter fault — and is used only when nothing better exists on that street, then flagged as such. Steeles Ave W at Shale Gate reads 1,823 vehicles a day beside a ten-day count of 19,725 on the next segment.
• School-zone speed sign — the nearest permanent Watch Your Speed radar sign within 400 m that has a monthly summary in the last 18 months. The City publishes vehicle counts in 5 km/h bins, so "over the limit" sums the bins above the posted limit and "10 km/h or more over" the bins from limit + 10 upward; the bin containing the limit itself is counted as at the limit. Months with fewer than 300 vehicles are ignored. A month whose median or 85th-percentile speed is zero, or with under 1,000 vehicles logged, is a sign that was not reading and is skipped; the latest valid month is used instead.
• Traffic calming — the Vision Zero unit's Traffic Calming Database, keyed by the same intersection IDs the centreline uses; a street segment is shown when either of its end intersections is this one. Planned installations are labelled as such.
• Ward and BIA — point-in-polygon against the 25 ward boundaries (with the sitting councillor from the same City file) and the Business Improvement Area boundaries.
• Restaurants and food — Toronto Public Health's DineSafe file lists every inspection of every food premises with the establishment's coordinates. It is a record of inspections, not a directory: it holds about two years of inspections (November 2023 to October 2026 at the last build), so a shop without an inspection in that window does not appear, and a premises that has since closed may. The file carries no establishment-type column, and Toronto Public Health inspects every kitchen in the city — daycares, care homes, hospices, school snack programs, stadium stands, office cafeterias, food factories and hot-dog carts along with the restaurants — so what a premises is comes from two screens. By name: an ordered word list files each premises as a restaurant or café, a grocery store, a specialist food shop (butchers, bakeries, fishmongers), a pharmacy, a convenience store or gas bar, or an institution; carts, stands, trucks and venue stalls are dropped, and so are numbered companies and names with a wholesale, factory or corporate word and no food word. By location: a premises still classed as a place to eat that stands within 60 m of a long-term care home, retirement home, hospital, licensed child-care centre or school and shares a distinctive word of its name with it is that institution's kitchen, and is dropped. A name is still not a category, so the odd premises is misfiled. Where the same business is registered more than once at the same address — including under the old and new establishment IDs the City introduced in 2026 — the record with the latest inspection is kept. Counts are within 300 m and 1 km; the five nearest restaurants and cafés are named. Inspection results are not shown: a count of infractions beside a name, without the inspection itself, is not a fact a reader can use; DineSafe's own site has the detail.
• Public washrooms and drinking water — three City lists merged: park washrooms (with the seasonal open/closed status and reason the City publishes), washrooms in City-operated buildings (civic centres, community centres, libraries where listed, with hours and an accessibility flag) and the automated street toilets. The nearest three within 1 km are shown, one per building. Drinking fountains and bottle fillers come from the parks water-sources list, within 600 m. Washrooms in malls, shops, restaurants and TTC stations are not in any of these lists.
• Child care — the City's licensed child care centres, within 1 km: licensed spaces by age group, whether the centre has a fee-subsidy contract with the City and whether it is enrolled in the Canada-Wide Early Learning and Child Care ($10-a-day) program. Spaces are licensed capacity, not vacancies; home child care is not included.
• Health and care — the nearest pharmacy, long-term care home, licensed retirement home and seniors' active living centre within 1 km, and the nearest hospital site within a 2 km walk, from the Ontario Ministry of Health's register of service provider locations (Open Government Licence – Ontario, refreshed yearly). The register lists eighteen kinds of facility and this site shows five. Mental-health and addiction organizations, HIV and hepatitis programmes, children's treatment centres and Indigenous health organizations are left out deliberately: a person using one has a privacy interest in it not being listed as a feature of a street corner. Private imaging and surgical centres are left out because the register names them after their licensee — a numbered company or an individual doctor. Primary-care practices (community health centres, family health teams, nurse practitioner-led clinics) are left out because they are rostered. Four hospitals are excluded by name on the privacy ground above. Hospitals are named by the name on the building rather than the register's corporate form ("Toronto Western Hospital", not "University Health Network - Toronto Western"). About a third of the sites are rehabilitation or specialty hospitals, and the register does not say which have an emergency department, so the page says "hospital site" and never "emergency"; beyond a 2 km walk it names nothing. The Ministry describes these locations as being for planning rather than navigation; they are used here to name a facility and say how far away it is, and for nothing else.
• Thefts at the corner — Toronto Police's Theft From Motor Vehicle and Bicycle Thefts files carry a location for every occurrence, but the police first move each one to the nearest intersection to protect victims' privacy. Because that is exactly our node set, an occurrence is joined to the node within 65 m of its published point — a match on the police's own key rather than a spatial guess. Counts use the same 2016–2025 window as collisions, with the most recent three years and the current year shown separately, and the two most common premises types. Other offences (auto theft, break and enter, robbery, assault) are published by neighbourhood only.
• Apartment buildings — the City's RentSafeTO register (rental buildings of three or more storeys and ten or more units), within 1 km. The roll is imported from LandlordReport, a sister site that keeps one page per registered building, so each building here links to the same page there and carries the counts it computes: the latest City evaluation score out of 100, final Landlord and Tenant Board orders naming the address in the latest half-year of Ontario's order catalogue, and open fire-code violations from Toronto Fire Services' inspection data. If that import is absent the build falls back to the City's registration and evaluation files directly (score and coordinates only). Buildings without coordinates in either file are left out; condominiums and houses are not registered and are not counted.
• Neighbourhood crime rates — Toronto Police's Neighbourhood Crime Rates: yearly counts and rates per 100,000 residents for nine categories by neighbourhood, using the City's population estimates. The site shows the latest complete year, the citywide rate for the same year (all neighbourhoods' counts over all neighbourhoods' population), the ratio between the two, and the change in the neighbourhood's own rate over five years. The "major crimes" figure sums the five Major Crime Indicators the police use — assault, auto theft, break and enter, robbery and theft over $5,000 — and the ranking is by that combined rate. Rates are per resident, which flatters residential neighbourhoods and penalises places with big daytime populations.
• Census profile — the City's Neighbourhood Profiles for the 2021 Census, 158-neighbourhood model: population, median age, median household income (2020), owner and renter households, dwelling types (apartments of five or more storeys, single-detached houses) and main mode of commuting. Density divides the census population by the area of the neighbourhood boundary polygon.
• Neighbourhood — point-in-polygon against the City's 158 neighbourhood boundaries; Neighbourhood Improvement Areas flagged from the City's classification. Neighbourhood KSI totals count every collision the City assigns to the neighbourhood, including those not matched to a counted intersection.

Layers are best-effort: if a source is unavailable when the site is rebuilt, that layer is left blank rather than stale.

How far is it?

Every distance on this site is the distance you would walk, not the straight line between two points. A shop 200 m away across a rail corridor is not a 200 m walk, and saying so would flatter every corner in the city that sits next to a barrier.

The walk is measured on the City's own Toronto Centreline — 59,033 segments of street, laneway, trail and walkway, 6,646 km in all, filtered to what a person can actually walk on. Expressways, ramps, rail corridors, rivers, hydro corridors and the shoreline are in the same City file and are excluded; ravine paths, park paths and the Waterfront Trail are kept, which is why a walk through a park counts. Streets the City lists as planned but not yet classified are left out, so the route can never take a shortcut down a street that may not exist.

The join is exact rather than approximate, which is unusual and worth stating. The centreline records the intersection id at each end of every segment, and those are the same City ids this site keys its pages on — so each of our 5,822 intersections is a node in that graph, matched by id. Nothing is snapped, so nothing is snapped to the wrong corner. Destinations do have to be attached to the network, and each is attached at up to four nearby doors with the shortest way in winning, because a park with two entrances has two ways in and a shop on a corner has more than one frontage.

How long that takes. Walking times beside the distances assume 5 km/h (1.4 m/s), the usual planning figure and the speed Google, Walk Score and transit apps assume, so the times read the way you expect. It is not the only defensible number: Toronto Transportation Services uses a slower 1.0 m/s in its own pedestrian network data, chosen to include older and slower walkers. If that is closer to your pace, add about 40%. Times are shown only for walks over 250 m, because below that the distance has already said it.

What we checked

Across more than 200,000 measured pairs, the typical walk came out at 1.25 times the straight line it replaced. Published figures for walking circuity in North American cities run from about 1.15 in a tight grid to 1.40 in sprawl, and the textbook factor for Canada is 1.30 ± 0.10, so ours sits where a city of Toronto's shape should. The neighbourhoods with the longest detours are Parkwoods-O'Connor Hills, Guildwood, Flemingdon Park, Don Valley Village and Elms-Old Rexdale — the parts of the city cut by ravines, rail corridors and the 401, which is the answer the geography predicts. The check script ships with the site.

What it still cannot see

• Crossing the road. Routes follow the middle of the street, not the pavement, so there is no cost to crossing and no notion of which side you are on.
• The PATH. Toronto's downtown walkway network has no open dataset, so it is not in the graph. Two downtown corners a long block apart may be a much shorter walk indoors than this site reports.
• Cut-throughs. Paths through a plaza, a parking lot or a school yard are not in the City's file unless the City drew them. Only 369 mid-block walkways are, which is certainly an undercount.
• Fine detail. Street shapes are stored simplified to about 10 m, so an individual distance can be out by a few metres either way. A walk is never reported as shorter than the straight line, because it cannot be.

Every one of these makes a walk look longer than it is, never shorter. Where this site is wrong about distance, it is wrong in the direction of understating how walkable a corner is.

"Within a walk" — our own score

Every intersection page carries a score out of 100 under Within a walk (until 16 September 2026 it was headed "Errands on foot", a name that promised more than City records can see: shops, banks, gyms and clinics are in no City dataset and score nothing anywhere), and the most walkable intersections page ranks the city by it. It is ours, computed here from the layers above — not a licensed walkability rating bought from anyone, and not comparable with one. It answers a single question: how much of an ordinary week is within a walk of this corner?

Six parts add up to 100. Each is a short ladder rather than a curve, so you can check the number against the lists on the page itself:

• Groceries and food — 30. Two or more grocery stores within 300 m score 15, one scores 13; failing that, two within 1 km score 6 and one scores 5. A grocery store means somewhere you could do a week's shopping — a corner store, a gas bar, a pharmacy or a dollar store is none of those and scores nothing here, which is a change from how this score was computed before September 2026. Places to eat within 300 m add 15 for fifteen or more, 13 for eight, 10 for four, 7 for two, 5 for one; failing that, 3 for ten or more within 1 km, 2 for three, 1 for one.
• Transit — 28. A subway, GO, UP Express or VIA station within 250 m scores 11; any station within 1 km instead scores 6. Surface stops within 300 m add 9 for six or more, 8 for four, 6 for two, 3 for one. Distinct routes within 1 km add 8 for twelve or more, 6 for eight, 4 for four, 2 for one.
• Library and places to go — 14. A Toronto Public Library branch scores 7 within 500 m, 5 within 1 km, 2 within 1.5 km. The City's Places of Interest and Attractions add 7 for three or more within 500 m, 5 for one; failing that, 3 for three or more within 1.5 km and 1 for one.
• Parks and recreation — 10. Parks within 500 m score 6 for three or more, 5 for two, 4 for one; a park only within 1.5 km scores 2. A community recreation centre or arena within 800 m adds 4; one only within 1.5 km adds 1.
• Crossing on foot — 10. A traffic signal at the node scores 4 (a pedestrian crossover, 3); a leading pedestrian interval adds 2, accessible pedestrian signals 2, traffic calming on an adjoining segment 2.
• Schools and child care — 8. Schools within 500 m score 4 for two or more, 3 for one; one only within 1.5 km scores 1. Licensed child care within 500 m adds 4 for two or more centres, 3 for one; one only within 1 km adds 1.

Highway ramps and collector-lane nodes are not scored or ranked. Until 16 September 2026 each score also carried a verdict phrase — "most errands need a car", "everything on foot" — in five bands: 85 and over, everything on foot; 73, most errands on foot; 52, some errands on foot; 38, a few errands on foot; below that, most errands need a car. The phrase is no longer shown: it was a judgement the records cannot support, since a corner with a plaza across the road in Vaughan, or a bank and a hardware store the City does not list, scored "most errands need a car". The number stays, with the city median beside it.

The score's thresholds are fitted to the real spread of scores, and re-fitted whenever the measurement underneath them changes. They moved once already: switching from straight-line to real walking distance took about nine points off every score.

What the score is measured against

There is no true walkability number to be right or wrong about, so the score is tested the only honest way available: against behaviour. The 2021 Census records how people in each Toronto neighbourhood get to work. Taking each neighbourhood's mean errands score and comparing it with the share of residents who walk to work gives a rank correlation of 0.73 across all 158 neighbourhoods — higher than population density alone manages (0.55) on the same test. The script that runs this check ships with the site, so the figure can be reproduced rather than taken on trust.

That test is also how the weights above were set. An earlier version counted parks, schools, libraries and child care anywhere within 1.5 km, and inside 1.5 km of a Toronto intersection there is almost always a park and a school: those parts were at full marks for about 70% of the city and were measuring nothing. Tightening the radii to 500 m and moving weight toward the parts that tracked behaviour raised the correlation and, more importantly, gave the score a usable spread.

One result is worth stating plainly rather than hiding. "Crossing on foot" correlates negatively with walking to work (−0.23 on its own). That is a confound, not a finding: traffic signals, accessible signals and leading pedestrian intervals get installed on arterial roads, and arterial roads are where people drive. It is kept in the score anyway, because whether you can actually cross the road is the one thing this site knows that a general walkability rating does not.

What it cannot see

Four limits, none of them papered over:

• Distances are walks, but not sidewalk-level ones. They follow the middle of the street rather than the pavement, so crossing the road costs nothing, and the PATH and informal cut-throughs are not in them. See “How far is it?” below.
• The shops it can see are food shops. The 30-point food part is built from DineSafe, which inspects food premises. A pharmacy, hardware store, bank or post office earns nothing, so a corner well served by those and short on restaurants scores lower than it deserves.
• Ladders have cliffs. A grocery at 299 m and one at 301 m score differently. That is the price of a rule you can check by eye against the lists on the page.
• The bands describe this site's own set of corners. Pages exist for intersections the City has run a traffic count at — about 5,838 of them — which is mostly arterials and commercial corners, not the quiet residential majority of Toronto's street network. Read a band as "compared with the other corners on this site", not "compared with every corner in the city".

It also measures what you can reach, not how pleasant or how safe the walk is. Collisions involving pedestrians are published on the same page and deliberately kept out of the score: a busy corner has both more shops and more collisions, and folding them together would mostly measure traffic volume.

What this site does not claim

• "Most dangerous" means most recorded serious-injury collisions. It is not a statement about cause, fault, road design or the behaviour of anyone involved.
• Counts reflect the day they were taken. Volume changes with construction, season and the count's own timing.
• No personal information is published. Person-level collision records are aggregated to counts; nothing on this site identifies an individual.

Licences

Contains information licensed under the Open Government Licence – Toronto. Collision data originates with Toronto Police Service and is published by the City under the same licence. Corrections and questions: see the data page.