Amtrak on-time performance · State Supported
Berkshire Flyer
Empire service. Customer on-time performance by fiscal quarter, FY22 Q4 to FY25 Q4, from the FRA's quarterly filings under 49 CFR part 273.
Not in the FRA's FY26 Q2 filing. Seasonal summer service; the FRA reports it only in the quarters it runs. The figures below are the quarters the FRA did report, ending FY25 Q4.
In FY25 Q4, 48% of Berkshire Flyer passengers got off the train within 15 minutes of the scheduled time, short of the 80 percent the federal rule asks for. The rule's standard is 80 percent in two consecutive quarters, which the last two reported quarters did not both reach.
FY25 Q4 is Jul 1 – Sep 30, 2025, the last quarter the FRA has published for this route, posted 2026-01-07. Federal record of past quarters, not a forecast.
Every quarter on record
0 of 8 reported quarters met the 80 percent metric.
Who the delay minutes were charged to, FY25 Q4
Most delay minutes on this route were charged to the host railroads that own and dispatch the track: 62% of the total.
| Responsibility | Delay minutes | Share | Per 10,000 train miles |
|---|---|---|---|
| Host railroad | 767 | 62% | 1,752 |
| Amtrak | 436 | 35% | 996 |
| Third party | 32 | 3% | 73 |
4,377 Amtrak train miles on the route in FY25 Q4, over 2 host railroads: Amtrak, Metro-North Railroad. Host-responsible covers freight interference, slow orders, signals and maintenance of way; Amtrak-responsible covers equipment, crews, station dwell and holding for connections; third party covers weather, trespassers, police activity, drawbridges and customs (49 CFR 273.3).
The quarterly record
| Quarter | Period | On time | Metric | Host /10K | Amtrak /10K | Third party /10K |
|---|---|---|---|---|---|---|
| FY25 Q4 | Jul 1 – Sep 30, 2025 | 48% | missed | 1,752 | 996 | 73 |
| FY25 Q3 | Apr 1 – Jun 30, 2025 | 59% | missed | 952 | 1,261 | 57 |
| FY24 Q4 | Jul 1 – Sep 30, 2024 | 74% | missed | 971 | 894 | 171 |
| FY24 Q3 | Apr 1 – Jun 30, 2024 | 51% | missed | 1,577 | 391 | 154 |
| FY24 Q1 | Oct 1 – Dec 31, 2023 | 33% | missed | 1,972 | 922 | 23 |
| FY23 Q4 | Jul 1 – Sep 30, 2023 | 50% | missed | 2,311 | 1,726 | 409 |
| FY23 Q3 | Apr 1 – Jun 30, 2023 | 66% | missed | 1,785 | 1,373 | 124 |
| FY22 Q4 | Jul 1 – Sep 30, 2022 | 51% | missed | 1,078 | 800 | 173 |
Informational only. The FRA's published figures for past quarters, not a prediction of any train and not live status. On time means a customer arrived no later than 15 minutes after the published scheduled arrival (49 CFR 273.5(a)).
Methodology and limits
Source: the Customer OTP Metrics, Delay Metrics and Delays per 10K TM Metric files the Federal Railroad Administration posts with each quarterly report under 49 CFR 273.5; this quarter's files are linked below. Customer on-time performance is the FRA's route-level figure, used unchanged, rounded to the whole percent here (the FRA's report floors it). It weights by passengers: a crowded late train counts more than an empty one. The 15-minute threshold is the rule's; it replaced Amtrak's older endpoint measure, which allowed a longer tolerance on longer routes, so figures before 2021 are not comparable. Delay minutes per 10,000 train miles are the route's minutes divided by its Amtrak train miles, times 10,000, by the party Amtrak coded the delay to; minutes on a host for which the FRA files no train miles count in the total but not the rate. Host railroads run most of the track outside the Northeast Corridor, which is why the split is shown. A quarter is a fiscal quarter: Q1 is October to December. One quarter can meet the metric without the route meeting the two-quarter standard. No station or train-level figures, no live status, no predictions.
- FY25 Q4 report (PDF)
- FY25 Q4 Customer OTP Metrics (xlsx)
- FY25 Q4 Delay Metrics (xlsx)
- FY25 Q4 Delays per 10K TM Metric (xlsx)
Last reviewed: October 2026
Related tools
On-time arrival rates, typical and worst-case delays, and cancellations for every airline on a U.S. route — from the DOT's on-time filings.
Open tool →Connection Risk CalculatorHow often a given layover at a connecting airport actually works, from the historical arrival record of the inbound route.
Open tool →