Best and worst times of day to catch a UK train
Published
Peter · Founder, Railly
Peter built Railly after one too many commutes spent guessing platforms and chasing Delay Repay claims. He verifies every guide against live timetable data, operator portals, and National Rail station records.
In short: Across the UK rail network, the worst hour for on-time trains is 08:00 (78.4% on-time). The best is 02:00 (97.5%). The 19-point gap is real and is driven by three compounding factors: junction congestion, depot availability, and passenger dwell time at peak.
Best and worst times of day to catch a UK train
A 24-hour curve of UK train reliability — what makes 08:00 worse than 11:00, and what makes 02:00 better than 22:00. The numbers below come from the same 30-day rolling window as our mainline operator ranking.
What "time of day" affects
Three things change between midnight and midnight on the UK rail network, and they all move reliability together:
- Train density. A peak-hour service runs every 2–4 minutes on busy commuter corridors. An off-peak service runs every 15–20 minutes. Higher density means higher junction contention and a smaller margin for recovery when something slips.
- Stock positioning. Early-morning trains are the first run of the day for that unit — if the depot was busy overnight, the unit might be in the wrong place, or briefly unavailable, or late onto the network. Trains mid-morning have had time to settle.
- Passenger loading. A commuter train at 08:30 has every seat full, doors closing fast, and dwell times at major stations of 30–60 seconds. An off-peak train has empty seats and can dwell for 20 seconds. Each extra 10 seconds at Clapham Junction or London Bridge cascades through the rest of the network.
None of these is hidden from the data. The 30-day on-time percentage traces them faithfully.
The full 24-hour curve
The chart above plots on-time percentage against hour-of-day across the whole network. The shape is the same every month, but the exact depth of the troughs shifts based on what's happening that month (engineering works, weather, industrial action).
A few key reference points:
| Hour | On-time | Notes |
|---|---|---|
| 02:00 | 97.5% | Best hour. Night engineering possession, but passenger services sparse and undisturbed |
| 05:00 | 93.1% | First commuter services start; stock positioning issues begin |
| 07:00 | 81.3% | First big peak trough; commuter loading + junction contention |
| 08:00 | 78.4% | Worst hour. Heaviest commuter density, first junction conflicts |
| 09:00 | 79.2% | Marginally better than 08:00; peak continues |
| 10:00 | 88.7% | Mid-morning recovery; stock settled, density dropping |
| 11:00 | 90.5% | Best of the working day — the sweet spot for off-peak travel |
| 16:00 | 84.2% | School-run + early commuter; first evening peak signals |
| 17:00 | 79.8% | Evening peak trough, but shallower than morning |
| 18:00 | 80.3% | Peak continues; the morning equivalent of 09:00 |
| 20:00 | 87.4% | Evening peak fades; reliability climbs back |
| 22:00 | 92.3% | Last long-distance services; very few conflicts |
The morning peak is the deeper trough because it starts from cold — overnight engineering possessions end at 04:30–05:00, and the first few hours of morning service are when the network shakes out the previous night's maintenance work.
Why peak is worse
The 11-point gap between 08:00 (78.4%) and 11:00 (90.5%) is structural, not random. The same train type, the same stock, the same depots — but the operating environment is fundamentally different.
The biggest single factor is junction contention at the London terminals. Clapham Junction, London Bridge, Waterloo, Euston, King's Cross, and Paddington each have a hard ceiling on throughput — around 22–24 trains per hour per pair of tracks. At 08:00, every operator is at the ceiling. One late arrival, one platform conflict, one signal stop, and the entire throughput collapses into a queue. The same junction at 11:00 is running at 60% capacity with plenty of slack.
The second factor is rolling-stock positioning. The 08:00 from Cambridge to London King's Cross is the first long-distance run of the day for that unit. It started at the depot at 06:30, ran ECS (empty coaching stock) to Cambridge, then formed the service. If anything went wrong overnight — a maintenance issue, a crew sickness, a cleaning overrun — the unit is late before it ever picks up passengers. The 11:00 service uses stock that has been running since 06:00 and is now in position.
The third factor is passenger loading. Heavy commuter loading at peak extends dwell times — a Class 700 at King's Cross might dwell for 50–60 seconds during peak, 20 seconds off-peak. Multiply that across 12 stops and the service is 6–8 minutes behind before it leaves London. The off-peak service doesn't have that load.
The TOC exception
The 24-hour curve is a network average, and it hides real variation between operators. Some TOCs are bad at every hour; some are surprisingly stable.
LNER, CrossCountry, and Avanti West Coast are bad all day. Their on-time percentage at 11:00 is still below the morning peak of better-performing operators. The cause is structural: long-distance services share junctions with freight, run on routes with few alternative paths, and depend on stock that's been on the network since 04:00. A CrossCountry journey from Plymouth to Edinburgh is more likely to be late than a ScotRail journey from Edinburgh to Glasgow, at any hour.
ScotRail, Merseyrail, Elizabeth Line, and Northern are reliable at most hours. Their networks are self-contained — they don't share junctions with long-distance or freight operators. ScotRail's 95.4% on-time (the best mainline operator) drops less than 5 points at peak. Merseyrail runs metro-style on a segregated network and barely moves.
Lumo and Hull Trains are open-access operators running on shared infrastructure. They suffer the same peak effects as the long-distance incumbents but at lower sample sizes, so their numbers move more.
If your route is on a TOC that holds up across the day (ScotRail, Merseyrail), the time-of-day effect is muted — your 08:00 is almost as reliable as your 11:00. If your route is on a TOC that's bad all day (LNER, CrossCountry), the time-of-day effect doesn't help much — your 11:00 is still bad, just less bad than your 08:00.
How to use this
Three practical takeaways for your next journey:
- If you have flexibility, shift to mid-morning or mid-afternoon. The 11:00 sweet spot averages 90% on-time; the 08:00 trough averages 78%. On a high-volume operator (LNER, CrossCountry), that 12-point gap is the difference between a Delay Repay claim being a rare event and a regular one.
- Don't fight the morning peak without a reason. If your day requires you to be at a 09:00 meeting across London, take the 07:30 rather than the 08:00 — the peak hasn't fully ramped up yet, and your 07:30 averages closer to 88–90% on-time than the 78% at 08:00.
- Watch the live board before you leave. The live boards on Railly carry a per-hour reliability chip on each row when the route has enough historical runs. The chip accounts for the time-of-day pattern, so an 08:00 row shows the 78% pattern and a 11:00 row shows the 90% pattern — they're not directly comparable without checking the hour.
For longer-distance operators where the time-of-day effect is muted, the operator ranking is the better guide. For specific routes, the live boards give the per-route per-hour breakdown that matters for your specific journey.
Methodology
The figures above are from the last 30 days of train running data on the National Rail network, aggregated by departure hour. We use the published scheduled departure minute from the timetable as the grouping key — same approach as our per-route reliability chip. A service that runs at 08:47 contributes to the 08:00 bucket; a service at 09:05 contributes to the 09:00 bucket.
- Window: rolling 30 days, ending yesterday.
- On-time: arrival at the destination within 15 minutes of schedule (the Delay Repay threshold).
- Sample size threshold: at least 1,000 runs per hour to be included. Quiet hours (overnight) sometimes fall below this and are excluded from the chart but appear in the table.
- Filter: passenger services only. Freight, empty stock, and charters excluded.
The colour bands in the chart above run green → amber → red by on-time percentage. Hours at 95%+ are green; hours below 85% are red.
Last verified: September 2026. Numbers come from the last 30 days of train running data on the National Rail network and refresh nightly. If you spot something that looks wrong, drop us a note.