Research · Evaluation · Updated Oct 6, 2026

Did bus cameras keep lanes clear? Bus lane cameras cut tickets 40% but brought no clear speed gain.

The MTA puts cameras on buses that ticket cars blocking the bus lane, and since 2024 cars at bus stops and double-parked. When the cameras reached a street that already had a lane, any speed gain was small: no more than about 0.6%. If a slow drift before the cameras had gone on, the data can’t rule out up to 3.3%, what a new lane gave. Tickets per bus-hour fell 40% in a year.

No step up when the cameras start

Bus speed on stretches with a bus lane, against similar stretches without cameras yet, by month before and after the cameras started. Weekdays, 6 am to 8 pm. The dashed line shows what a new lane gave, for scale.

New York’s bus lane cameras brought no measurable speed gain

Weekday bus speed, lane stretches with cameras minus similar ones without yet, by month from camera start, 2023–26, %, with likely ranges (95%)

Source: MTA bus segment speeds; MTA bus camera routes; NYC DOT bus lanes

Each month is measured against the six months before the cameras started, so those months average zero by design. The start month is left out.

Tickets fell 40% in the first year

Camera tickets per bus-hour on each route, by month after the cameras started, as a share of the route’s first three months.

Tickets from New York’s bus lane cameras fell over their first year

Camera tickets per bus-hour on each route, by month after its cameras started, 2022–26, % of the route’s first three months

Source: MTA bus camera violations; MTA bus route speeds

No check finds a clear gain

Each line is a separate comparison of streets with cameras against streets without them yet, with its 95% interval. None finds a clear gain. The largest estimates, about +1.5%, are half of what a new lane gave, and their intervals include zero. Splits by borough and by start month are in the page’s data, not here: with 14 such splits (five boroughs, nine start months) a few will be clear of zero by chance, and five are, in both directions.

Speed change once cameras start

    What could change this answer

    The drift before the cameras is the one that matters most: allow for it and the best guess rises from −0.2% to +1.5%. The rest mostly say what this measures, not how big it is.

    • How big a gain we could have seen (either way)The plain range, −1.1% to +0.6%, rules out any gain above 0.6%: about a sixth of the 3.3% a new lane gave (0.6 ÷ 3.3).
    • A drift before the cameras (would raise it)Camera streets were slowly losing ground in the year before. If that had gone on, the cameras offset it, and the gain is +1.5% (−0.2 to +3.3). Most of the drift is in months 12 to 10 before, so we treat this as the upper case.
    • Thin coverage (would hide a gain)The cameras ride on some buses, not all, and only ticket when two camera buses see the same car, so a street gets far less coverage than from a fixed camera.
    • Enforcement already there (would hide a gain)Most lanes here already had NYPD enforcement and DOT’s fixed cameras. This measures what the bus cameras added to that, not what enforcement is worth in all.
    • Tickets aren’t blocking (would shrink it)Tickets per bus-hour also fall when fewer camera buses run on a route, so the 40% may overstate how much less drivers block the lane.

    What it means

    • What the data showOn streets that already had a bus lane, bus cameras added no more than about 0.6% to bus speeds, or at most about 3% in the upper case. A new lane gave 3.3%. Camera tickets per bus-hour fell 40% within a year.
    • What follows, if it holdsAdding cameras to a street that has a lane won’t, alone, make its buses much faster; a lane on a street without one does more. The cameras may still pay off in ways we didn’t measure: how often lanes are blocked, how even the service is, and crashes. These data can’t say whether the program is worth its cost.

    How this works

    The cameras reached routes in batches: a few in 2019–20, more in late 2022 and 2023, then dozens from September 2024 on. Each batch is compared with streets whose buses had no cameras yet, over the same months. Streets that had cameras before the comparison began never serve as the comparison.

    • StretchesThe MTA’s bus speeds between timepoint stops, by month, January 2023 to August 2026. A stretch counts as under cameras from the month camera routes cover half its path, so a local bus on a street where an SBS route has cameras counts too: the cameras ticket the cars in its way as well.
    • LanesThe main comparison uses stretches with half or more of the path on a bus lane that opened at least seven months before the cameras, and no new lane in the year after. That isolates the cameras from the paint.
    • The comparisonFor each start month, each stretch’s speed against its own average in the six months before, minus the same for stretches with no cameras yet, matched on borough (Manhattan split at 60 St), express or local, and baseline speed. Weighted by bus-miles. Effect = the average of the 12 months after.
    • Intervals95% intervals from resampling whole routes 1,000 times.
    • The checkIf camera streets were already pulling ahead or falling behind, it would show in the year before. They were slowly falling behind: months 12 to 7 before came out +1.3% against the six months just before, about 0.17 points a month. If that drift had gone on, the cameras would have had to offset it, and the gain would be +1.5% (95% interval −0.2 to +3.3) instead. The whole-route comparison has no such drift and finds no gain.
    • TicketsCamera violations by route and month (MTA), all statuses except those rejected for technical faults, per 100 bus-hours from the MTA’s route data. Routes with tickets in their first three months; the latest months are still filling in and are left out.
    • How sure we areWe grade each answer high, medium or low. High: the range is narrow and every check agrees. Medium: the answer holds, but its size depends on a choice we made, or a check disagrees. Low: the answer rests mainly on an assumption we can’t test.

    Speed includes time at stops and lights. The caveats that matter most are under What could change this answer.

    What to watch next

    Cameras keep reaching new routes: 21 so far in 2026. Each batch adds a year of before and after, which will narrow the range and show whether the drift before the cameras was real. Watch, too, whether tickets keep falling or level off. Every number here is in bus-cameras.json, so you can check our work.