How Train Delays Are Measured (and Why 'On Time' Differs by Country)

01

'On time' is a definition, not a fact

Compare two trains. An ICE reaches Frankfurt five minutes and 59 seconds behind schedule, and Deutsche Bahn logs it as punctual. A Tokaido Shinkansen arrives 90 seconds late into Nagoya, and JR Central's control room records a delay. Same physics, opposite verdicts, because punctuality is whatever a railway defines it to be. Three choices shape every national statistic: the threshold (how many minutes count as late), the measuring point (final destination only, or every station), and the population (do cancelled trains count at all). Germany uses under six minutes at measured stops. Switzerland uses three. Britain's headline measure now uses 59 seconds. Japan's operators track to the second and issue delay certificates from five minutes. Amtrak is judged on whether customers reach their own station within 15 minutes. None of these numbers can be compared directly, which is exactly why operators quote them so freely. Once you know the definitions, league tables stop fooling you.

02

Japan: punctuality measured in seconds

Japanese railways treat the timetable as a promise, not an aspiration. JR Central reports the average delay per Tokaido Shinkansen train each year, and the figure routinely lands under two minutes, a number that includes typhoons and earthquakes across more than 350 daily departures. Delays are logged at every stop; a driver arriving 15 seconds down will note it. From five minutes, staff issue chien shomeisho, small delay certificates you hand to your employer or school, because a five-minute delay is considered disruptive enough to need proof. The culture cuts both ways: in 2017 a Tokyo-area operator made international news by apologising after a train left 20 seconds early. Structural choices make the precision possible: dedicated high-speed tracks with no freight, standardised rolling stock, and dwell times planned to 15-second granularity. The lesson for travellers is simple. In Japan, plan connections of five minutes with confidence. Nowhere else should you dare.

03

Germany's 5:59 rule and what it hides

Deutsche Bahn counts a train as punctual if it arrives less than six minutes late, the so-called 5:59 rule. Even by that generous yardstick, 2025 was grim: 60.1 per cent of long-distance trains ran punctually, down from 62.5 in 2024 and the worst annual figure in decades. Regional services managed 88.7 per cent. Two accounting choices flatter these numbers further. Cancelled trains vanish from the statistic entirely, so a service axed at short notice damages the figures less than one crawling in 20 minutes down. And skipping stops to recover time can launder a delay from the record. DB also publishes a more honest metric, passenger punctuality, which counts travellers reaching their final destination within 15 minutes, missed connections included; that stood at 65.5 per cent in 2025, down from 67.4. The cause is a network being rebuilt while in service after decades of underinvestment, with entire corridors such as Hamburg to Berlin closed for months of reconstruction. Pad any German connection by 20 minutes, and never book the last train of the evening.

04

Britain and America: minute-counting versus endpoint games

Britain historically used the Public Performance Measure, which forgave five minutes at the final destination for commuter operators and ten for long-distance ones. Since 2019 the headline metric is On Time: the percentage of recorded station stops made within 59 seconds of schedule, which is why UK punctuality suddenly looks worse than its neighbours' despite measuring more honestly. Roughly seven in ten British station calls meet that standard. The United States runs the opposite philosophy. Amtrak was long judged only at the endpoint, and timetables carry generous recovery padding before major terminals, so a train can lose an hour en route and still arrive officially on time. Federal rules from 2020 introduced a fairer customer on-time metric: 80 per cent of passengers should reach their own station within 15 minutes. Long-distance routes regularly miss it, largely because Amtrak runs on freight-owned tracks where dispatchers control priority. Wherever you travel, learn which game the operator is playing before trusting its scorecard.

05

How one late train becomes fifty

Delays multiply because trains, crews and platforms are booked back to back all day. A carriage set finishing one journey immediately forms the next; planners call this the diagram. When the 08:10 arrives 25 minutes down, the 08:45 it was due to form leaves late too, and so does everything that set touches until midnight or a gap lets it recover. Crews cascade the same way, with legal rest rules turning a modest delay into a cancellation when a driver times out. Then comes the network effect: a late train misses its planned slot through a junction or single-track section and must wait for the next one, delaying itself further and everything queued behind it. British data attributes roughly 60 per cent of delay minutes to this knock-on category rather than to the original incident. Operators fight back with recovery margin in schedules, spare units and standby crews, but on a congested network the arithmetic is unforgiving. This is why a 10-minute fault at 07:00 still echoes at 19:00.

06

Claiming compensation: know your thresholds

Delay pay is rule-based, so learn the numbers. Across the EU, Regulation 2021/782 guarantees 25 per cent of your ticket price back for a delay of 60 to 119 minutes at your final destination and 50 per cent from 120 minutes, though operators may refuse payouts for extreme weather and similar force majeure, an exemption added in 2023. Deutsche Bahn administers this as Fahrgastrechte; claim in the app or with a form, and payouts under €4 are not made. Britain's Delay Repay is more generous: most operators pay 25 per cent of the single fare from just 15 minutes, 50 per cent from 30, 100 per cent from an hour, and 100 per cent of a return from two hours. Japan refunds the limited express supplement, not the base fare, once a train runs two hours late. Amtrak offers no statutory scheme, only goodwill vouchers. Everywhere, claims need your ticket and usually arrive within a month. File every time; the thresholds exist because passengers use them.

07

Why the statistics mislead, and how live trackers estimate delays

Treat any punctuality percentage with suspicion until you know three things: the threshold, the measuring point and what happened to cancelled trains. A railway that pads its timetable, that measures only at the terminus, or that quietly removes cancellations can report a rosy figure while its passengers seethe. Right-time performance at every stop, the standard Japan and now Britain apply, is the only measure that reflects what you experience mid-journey. Live tracking works from different raw material altogether. Signalling systems report when a train occupies and clears each track section; operators feed this into prediction engines, such as Darwin in Britain, which compare actual against planned running and project the arrival forward, station by station. That is why an app's estimate jumps: each junction passed either confirms or rewrites the projection. The forecasts trust scheduled recovery margin, so predictions often improve as a train runs. When an estimate keeps creeping upward in small steps, assume the problem is still live and plan around the worst number you have seen.

FAQ

Frequently Asked Questions