Skip to content
Docs

Subway slow zones, 2022–2026

Checked against news: Slow zones: checks against the news. Data: subway feed archive, weekdays April 2021 – October 2, 2026. Run 2026-10-04.

Why

Section titled “Why”

Boston’s TransitMatters slow-zone tracker pushed the MBTA into publishing its own slow-zone data and fixing them. The MTA publishes no list of NYC speed restrictions. This is a first version of one.

Method

Section titled “Method”
  1. Segment times. For each train, the time from leaving one station to leaving the next (marked_past; includes dwell at the second). This is 1,263 directional segments.
  2. Daily figure. Weekday trains 6:00–21:59, trimmed mean (10th–90th percentile). A median would jump in 15 s steps, because times arrive at the feed’s refresh rate.
  3. Baseline. 10th percentile of the segment’s daily figure over the previous 120 weekdays: a good typical day, which adapts only to changes lasting months.
  4. Slow day. At least 10% and 10 s over baseline, with 20+ trains. Slow zone: 4+ slow weekdays in a row (one gap day allowed).
  5. Kinds. Slow zone: median extra time up to 2 min and 150%. Disruption: more than that. System-wide change: 15+ zones starting the same day (timetables, network-wide events). Service change: the lines using the segment differ from the 60 weekdays before, or the segment is new (route swaps, reroutes, reopened stations). Checking against the news showed these are common.
  6. Cost. Extra seconds × trains, as train-hours; and extra seconds × riders passing through, as rider-hours. Riders come from routing every station-to-station trip in the MTA’s 2025 origin-destination estimates (3.66M weekday daytime trips) along its fastest path.
  7. Reported from 2022. Baselines in 2021 come from near-empty pandemic trains, so the 2021 ridership return looked like hundreds of slow zones.

Results

Section titled “Results”

Weekly rider-hours lost to slow zones

  • Slow zones cost riders about 2 million hours a year (weekdays 6 am–10 pm): 1.87M in 2022, 1.98M in 2023, 2.20M in 2024, 1.98M in 2025. Right now it’s about 8,600 rider-hours each weekday across 109 active zones.
  • 7,635 slow zones since 2022; 603 cost 25+ train-hours each. A further 1,068 were set aside as service changes and 1,653 as system-wide changes.
  • Biggest since 2022: E Roosevelt Av → Queens Plaza (Apr–Jul 2023, +50 s on ~141k riders a day, 124k rider-hours); F through the 63 St tunnel (Aug 2023–Jan 2024, 110k); 6th Av express 42 St → 47–50 Sts (Aug 2023–Jan 2024, 88k).
  • Active now, top 3: N/W Queensboro Plaza → 60 St tunnel (+50 s since Jun 8), R Queens Plaza → 60 St tunnel (+62 s since Jun 1), 7 40 St → 46 St (+34 s since May 25).

Validation

Section titled “Validation”

Searching news and MTA notices for 13 detected zones found a matching cause for 7, often to the day (details):

  • Real slowdowns: the 7 at 40–46 St (temporary platforms from May 25, 2026, same day); Broadway–Lafayette → W 4 St (switch work from May 15, 2023, same day; the MTA warned of ~2 extra minutes).
  • Service changes, now filtered out: the M into the 63 St tunnel (F/M swap, Dec 8, 2025); Clark St (station reopened May 4, 2022, so trains stop again).
  • Work-related congestion, still counted: 6th Av at 42 St during the 63 St tunnel work. Riders really lost that time, but it wasn’t a speed restriction.
  • No source found: the N/R/W 60 St tunnel zones (June 2026), the D at Norwood, the 6 at Grand Central (2022).

Caveats

Section titled “Caveats”
  • Run time includes dwell. A segment can look slow because platforms got busier, not because of a speed restriction. A step change is the tell; slow drift is less clear-cut.
  • “Track point” segments (e.g. R60, the 60 St tunnel) are timing points the feed reports between stations.
  • One stretch of slow track shows up as several zones, one per segment and direction. These aren’t merged yet.
  • Same-day bursts are set aside, but some may be real, for example a timetable change that adds running time. (The N/R/W Lexington Av/59 St zone from June 29, 2026 was set aside this way.)
  • Rider counts are 2025 averages, routed by fastest path with no waiting time, which sends too many riders to express lines.
  • Some “slow zones” are work-related congestion (extra trains merging, terminals under construction) rather than speed restrictions. The page says “running slower than usual”, not “speed restriction”.
  • Not checked against MTA records of actual speed restrictions, which aren’t published. Confirming a sample of zones with riders, news reports or MTA board books would strengthen it.