All Through the Town

The latest comparable week

—

Loading…

Speeds are local, limited and Select Bus Service routes, each counting once, with time stopped included and every hour from 6 a.m. to 11 p.m. counting equally. Weekday figures use the latest week whose weekdays were fully caught; when one or two hours were missed, each is assumed to have moved like the rest of that week’s day (the notes in the table below list them). Each figure is set against the average of the four comparable weeks before it. A move smaller than that series’ usual week-to-week wobble is labeled “no real change.” The faint line is every comparable week since collection began; the bold line is its straight-line trend.

Speed over time, by borough

Weekday speed with stops for each borough’s local, limited and Select Bus Service routes, plus its two busiest routes. Comparable weeks only. Click a route for its full history.

Watch routes, fixed so the series stays comparable: Manhattan M15+ and M4, the Bronx BX12+ and BX36, Brooklyn B6 and B41, Queens Q44+ and Q58, Staten Island S79+ and S53 (“+” marks Select Bus Service). The figure under each is the net change its trend line implies across the comparable weeks shown. Every other route is on its own page.

Boardings, by route

Average weekday boardings as estimated by the MTA from the passenger counters on its buses (MTA bus stop ridership estimates, published monthly about a month behind), beside paid fare taps (OMNY and MetroCard). Explore stop by stop on The People on the Bus.

Loading…

Boardings here are the MTA’s estimates: it reads the passenger counters that some buses carry and scales them up to stand for all boardings. Boardings are not people: a rider who transfers boards twice. Fare taps count only paid entries (OMNY and MetroCard), including free transfers. The estimates run well above taps: systemwide by 78% to 85% from June to August 2026, and on the typical local route by about 77% in August. These two independently constructed series differ substantially. Their ratio is not an estimate of fare evasion. For evasion, use the MTA’s separately published measure, which put evasion on local and limited buses at 48.7% in the third quarter of 2026 and on express buses at 4.8%. Until October 2026 the MTA published unscaled counter totals instead, which ran only slightly above taps; it replaced the whole series, and every month shown uses the new estimates. The MTA now reports the Q52 SBS and Q53 SBS together, so both show the pair’s combined figures. If buses go fare-free, taps stop and the estimates are the only route-level count left, which is why the ratio is tracked here now. Each change compares the latest three months two ways: with the same three months a year earlier, which is fair to the season, and with the twelve months before, which shows the recent run. “Route changed” marks a route redrawn by the Queens bus network redesign (June 29 and Aug. 31, 2025) inside that window.

Geography cuts

Bus priority corridors: routes whose shape runs at least 30% of its length within about 60 meters of a marked bus lane or busway (New York City Department of Transportation bus lane data), against routes with little or none (10% or less). Congestion zone: routes the MTA lists as operating in or crossing the Central Business District, against the rest. Each figure is the plain mean of member routes’ weekday speed with stops. Compare each group with its own past: bus lanes go where traffic is worst, so a gap between groups says more about geography than about the lanes.

Week by week

Every week since collection began. Grey rows are not comparable; the note says why.

ISO week	Dates	Hours caught	Comparable	Weekday	7–10 a.m.	4–7 p.m.	Stopped	Weekend	Moving only	Buses	Scheduled	Notes
Loading…

What else happened

Holidays and service cuts, collection gaps, rain and snow at the Central Park weather station, and changes to how the tracker works, over the last ten weeks. The same markers appear on the charts.

Download the data

Every series behind this page as flat CSVs, regenerated with each data refresh. Free to use with attribution. Read the data dictionary first.

Daily, whole system one row per day, with hours caught and holiday flags Weekly, whole system weekday, weekend, rush hours, buses vs. schedule Monthly, whole system same columns by calendar month Weekly by borough weekday and weekend speed per group Weekly by route every route, every week, with comparability flags Monthly by route every route, every month Riders by route, monthly MTA estimates since January 2024, beside fare taps Tracker vs. MTA monthly, built the same way Events holidays, gaps, weather, method changes Stop-level boardings 13,351 stops × hour of day, June 2026 Route geography classes bus-lane share and congestion-zone relation Data dictionary every column defined, and what not to do with it Method changelog every change to collection or computation, dated

Speeds are our own estimates from GPS positions, not official MTA metrics; ridership is the MTA’s own counter and fare data. Raw JSON with more fields lives in data/summary/, data/route-history/ and data/daily/ in the repo; raw snapshots are on its bus-data branch.

Data collection — how it runs and what to be skeptical of

How collection runs. One long-running GitHub Actions job polls the MTA BusTime feed about a dozen times over six minutes at 9 and 39 minutes past every hour, 6 a.m. to 11 p.m. Eastern, and saves every bus the feed reports in service. Before its six-hour time limit it starts its own successor. Until October 2026 each burst was a separately scheduled run, and GitHub skipped many of them; from Oct. 1 to Oct. 6, 2026, only three to five bursts a day were caught. The bar strip at the top of this tray shows the last three weeks, and the daily processor fails loudly when a day comes in under 12 of 17 hours. No bus position is interpolated. Weekly figures may fill in up to two missed hours per day type, as the methodology describes, and the weekly table lists which.

How speed is computed (collector/process.js and collector/rollup.js, method version 2). See the methodology note in the sidebar and the data dictionary. Every day since April 22, 2026, was reprocessed from the raw snapshots when the method changed, so the whole series is computed one way.

Known limits.

Straight-line distance over half a minute slightly undercounts a bus turning a corner.
Collection caught only some hours of most days before late June 2026, so few spring weeks are comparable.
Staten Island express routes read slower than the MTA’s figures, for reasons not yet known. Express routes are kept out of the index.
Scheduled trips come from the MTA’s published timetable (GTFS) and exist from Sept. 6, 2026. Across September 2026 the tracker saw 96% to 98% as many buses on the road as trips scheduled to be under way; a trip not run, or a bus whose GPS isn’t reporting, pulls that down.

Built by Josh Greenman.

Speed with stops

Right now

Week by week

Checked against the MTA

What this page no longer shows

General caveats