This is the data behind Thatcham’s trains: ready for 2,500 more homes?. For the short version, start there.

Data and definitions

• Sources: Realtime Trains (every train calling at the nine stations from Newbury to London Paddington, from 8 July 2026); the Office of Rail and Road’s passenger rail performance statistics; Network Rail’s delay attribution, via the Rail Data Marketplace (from February 2026).
• Journey: a trip on a single train between Thatcham and Newbury, Reading or London Paddington, in either direction, as timetabled on the day. Rail replacement buses are excluded.
• On time: arrived less than 5 minutes late - the railway’s standard Public Performance Measure (PPM) for London & South East and regional trains, and ORR’s “Time to 5”. 15 minutes is also shown, as that’s where Delay Repay compensation starts.
• Failed: the train was cancelled at either end, including trains that skipped Thatcham or ended their journey early.
• Dates: figures quoted in the text cover 8 July to 9 November 2026 unless stated. The charts and tables update daily.

Getting to the station

Straight-line distances from the site to Thatcham station (method):

Walking routes along real roads and footpaths (OpenRouteService). The part of each walk inside the site is a straight line, as there are no street plans yet, so these are best cases:

Conclusion: most of the homes would be in the north and east, 30 to 40 minutes’ walk from the station, so most residents who use the railway are likely to drive to it. The station has 55 car parking spaces, most of them on the far side of the level crossing from the town.

The level crossing

• Traffic: around 9,800 vehicles and 450 pedestrians on an average weekday, with around 140 trains (2023). Network Rail puts peak traffic at 300 to 400 vehicles an hour (Newbury Today).
• Waits: an average of 10 to 15 minutes has been reported, up to half an hour at the back of the queue (Newbury Today, February and July 2026).
• Why it’s slow: the signal protecting the crossing for trains towards Reading is over a mile away, so the barriers come down long before a train arrives (Newbury Today).

Why not just build a bridge?

• At the station: the railway, the River Kennet and the Kennet & Avon Canal run close together, with buildings alongside, and flood risk has always made a site hard to find. In 2015 Network Rail and the council put the cost at over £20m. Talks under the previous MP, Laura Farris, didn’t go ahead. A 2021 review for the council’s Thatcham Strategic Growth Study looked at the options again.
• At Colthrop: in 2019 the Colthrop Village Consortium proposed 950 homes south of the railway and canal in the Local Plan Review, with a road bridge from a new roundabout on Pipers Way to Crookham Hill so the crossing could close. It estimated the bridge at about £12m (more recently £16.5m), paid for by the housing and built before any homes.
• Network Rail’s view (2021): “We support the closure of Thatcham level crossing, therefore fully support any development that includes mitigation in the form a bridge that can replace the existing road” (Newbury Today).
• The council’s concern: its planning policy manager said a bridge would “become a rat run to go into Basingstoke and Greenham” (same article). Residents of Crookham Hill and Thornford Road raised the same worry.
• Outcome: the council allocated North East Thatcham, with no bridge. The Colthrop scheme was later combined with Henwick Park as an alternative strategy, which wasn’t taken up.

What’s happening about it

• February 2026: Newbury MP Lee Dillon tabled amendments to the Railways Bill requiring Great British Railways to review high-delay crossings like Thatcham every year and “seriously consider” replacing them. The outcome isn’t known.
• June 2026: Network Rail is looking at electronic warning signs, which would warn drivers but not reduce how long the barriers are down. No decision yet.
• July 2026: a Thatcham resident built a live tracker to predict when the barriers will drop.
• November 2026: residents are still asking about a bridge or other solution.

And the new development?

The council’s masterplan FAQ (4.5) says the baseline transport work “did not consider the impact of the site on the Level Crossing to be significant above the existing situation”, that a bridge “would be technically challenging and costly to deliver” and “would need to cross the river and canal”, and that the crossing will be addressed “outside of the masterplanning process”.

Conclusion: the crossing is outside the masterplan. The FAQ’s objections describe a bridge at the station, not a crossing further east like the one proposed at Colthrop.

How reliable are the trains?

Key results

What works well

• Most journeys are close to time: 75% to 88% of journeys to and from Newbury and Reading arrive within 5 minutes, and 93% to 96% within 15 minutes.
• Most trains run. About 3% of calls at Thatcham are cancelled, lower than GWR as a whole.
• Direct trains to London are quick - 42 to 46 minutes to Paddington, timetabled.
• Saturdays are good - 77% to 92% of journeys arrive within 5 minutes, depending on the destination.
• Trains towards London rarely run short of coaches (2% to 4% in the morning peak).

Where the problems are

• London journeys are the least reliable: 65% to 70% arrive within 5 minutes (87% within 15), and 7% to 8% fail outright. A daily commuter would be 5 or more minutes late, or not get there, about one day in three.
• About 1 in 6 regular direct London trains didn’t run direct. Most were cut back to Reading or dropped from the timetable in advance, which doesn’t count as a cancellation. September was the worst month.
• The evening journey home is the weakest. From London between 16:00 and 19:00, 12% of direct trains to Thatcham failed and 59% arrived within 5 minutes.
• The last weekday train from Reading didn’t run on about 1 in 11 evenings.
• Weekdays are worse than weekends, there are no direct London trains on Sundays, and replacement buses ran on 7 of the 18 Sundays.
• Morning trains towards Newbury are short of coaches about 1 in 10 times.
• About two thirds of cancellations are down to Network Rail’s infrastructure and about 30% to GWR; delays on the line through Thatcham nearly doubled between February and August 2026, mostly knock-on from elsewhere.
• GWR ranks 16th of 23 operators nationally; the Elizabeth line, on the same tracks between Reading and London, does much better.
• A daily London commuter would be owed about 12 single fares a year in Delay Repay in each direction.

Journeys, by destination

Weekday journeys only.

Results

• The story is different for each destination. Newbury to Thatcham is the most reliable (88% within 5 minutes). Journeys to and from London are the least reliable: 65% to 70% within 5 minutes, and about 7% to 8% failed outright.
• Most delays are small. Within 15 minutes, every journey looks much better: 93% to 96% for Newbury and Reading, 87% for London. What sets London apart is mainly the higher failure rate.
• For a daily London commuter: about 80 journeys a year (one in three) would be 5 or more minutes late or fail, including about 15 that fail outright.
• Reading to Thatcham is worse than Thatcham to Reading (75% against 82% within 5 minutes). Coming home is worse than going out.

What Delay Repay would pay back

What a regular traveller could claim. GWR has been in the Delay Repay 15 scheme since 2019. The minimum rates for a single ticket are 25% of the fare for a delay of 15 to 29 minutes, 50% for 30 to 59 minutes, and 100% for an hour or more (Transport Focus).

* If you made this journey once every working day for a year (230 journeys). This is an illustration, not a claim calculator: it uses lateness at the destination as the delay, assumes a cancelled train costs one wait for the next train (about 41 minutes on average at Thatcham, so 50%), and season tickets are compensated differently.

Results

• A daily commuter to London could claim on about 1 in 8 journeys, and would be due back the equivalent of about 12 single fares a year in each direction - roughly 24 single fares for a year of daily return trips.
• Local journeys would get back about 4 to 6 single fares a year in each direction.
• Conclusion: by the railway’s own compensation rules, about 5% of the value of a London ticket is owed back.

Direct London trains that didn’t run direct

Official figures only count a train as cancelled if it’s cancelled on the day. Trains cut back to Reading or removed from the timetable in advance don’t count - for example, the 09:25 from Thatcham to Paddington (train W35113) was planned to terminate at Reading every day from 7 to 11 September 2026.

Method: a “regular direct train” is a timetabled train (one schedule ID) that ran direct between Thatcham and Paddington on at least 80% of the days it appears, and at least 10 times. It is expected on the same days of the week between the first and last time it appears. “Not in the timetable” is inferred - a timetable change that gives a train a new ID looks the same - so treat it as approximate.

Results

• Across July to early November, about 1 in 6 expected direct London trains (533 of 3,390 runs, both directions) didn’t run direct. Only around 3% were “cancelled at Thatcham” - the only part that would appear in official-style figures.
• September was the worst month: more than one in five direct trains was planned to run short, and only about 70% ran direct in either direction.
• August had the most trains that set off but didn’t make it to London (about 6% to 7%).

Day of the week

Results

• Saturdays are the best day for every journey - between 3 and 18 percentage points better than weekdays. The service is at its best when commuters need it least.
• Weekdays are consistently middling. Mondays have the highest failure rate for London journeys (8.6% of trains from Thatcham to London failed).
• Sundays are poor for Reading and Newbury (around 70% to 77% within 5 minutes), and there are no direct trains between Thatcham and London on Sundays - every Sunday London journey means changing at Reading.

Time of day

“To” means a journey from Thatcham; “From” means a journey back to Thatcham. Times are when the journey starts.

Results

• Mornings are good. Thatcham to Reading before 10:00 arrives within 5 minutes 85% to 88% of the time.
• The evening journey home is the weakest. From London between 16:00 and 19:00, 12% of direct trains to Thatcham failed and only 59% arrived within 5 minutes. From Reading between 16:00 and 21:00, 60% to 65% arrived within 5 minutes.
• Late at night, more trains are cancelled. After 21:00, about 6% of trains from Reading to Thatcham failed, and 8% of direct trains from Thatcham to London.
• There are almost no direct trains from Thatcham to London between 19:00 and 21:00 (so that slot is left out of the chart), and very few direct trains home from London after 21:00 (39 in four months of weekdays). Late journeys mostly mean changing at Reading - and these figures can’t show missed connections there.

The last train home

Results

• The last weekday train from Reading to Thatcham (usually around 23:00) didn’t run on 8 of 86 evenings - about 1 in 11. When that happens, the realistic option is often a taxi from Reading, which costs around £50 to £60.
• The last direct train from London (around 20:00 on weekdays) didn’t run on 5 of 86 evenings.
• At weekends the last train from Reading ran more reliably, but was 15 or more minutes late on about 1 in 7 evenings.

Short trains

Trains running with fewer coaches than they usually have.

Results

• Morning trains towards Newbury (07:00 to 10:00) are short-formed about 1 in 10 times - the highest rate of any time or direction.
• Evening peak trains towards Newbury (16:00 to 19:00) are short about 1 in 18 times; morning trains towards Reading and London are short less often (2% to 4%).

Short-formed is measured against each train’s own usual number of coaches, so a train that is always too short isn’t counted. Weekdays only.

Disruption calendar

Each square is one day: the share of scheduled calls at Thatcham cancelled or replaced by a bus.

Results

• 22 of the 122 days had some rail replacement buses, including 7 of the 18 Sundays. On Sunday 27 September, 20 of 33 calls at Thatcham were buses.
• The worst weekday was Monday 14 September: 21 trains cancelled and 5 buses, a third of the day’s service.
• Planned engineering closures that remove trains from the timetable entirely don’t show up here.

Why trains are cancelled or late

Network Rail’s delay attribution (Rail Data Marketplace), February to August 2026: the cause of each delay and cancellation, and who was responsible - Network Rail (infrastructure), the train operator (trains and crew) or someone else. Cancellation figures cover GWR stopping trains that start or end at Newbury or Bedwyn, which make most of the calls at Thatcham.

Most common specific reasons:

Per four-week period: cancellations of these trains, and delay minutes on the line through Thatcham (Newbury to Southcote Junction, just west of Reading).

Results

• About two thirds of cancellations were down to Network Rail - mainly infrastructure faults such as points and axle counter failures, overhead line faults and track problems, plus trespass and other external events. About 30% were down to GWR, mostly train faults (18%) and crew (7%).
• About 1 in 5 cancellations is recorded as “uninvestigated” - the cause was never established.
• About a third of cancellations were partial - the train ran for only part of its journey, ending early or starting somewhere other than planned.
• Delay minutes through Thatcham nearly doubled between February and August 2026 (from about 1,750 to about 3,300 per period), and most of it (52% to 84%) was knock-on delay from incidents elsewhere, rather than problems on this stretch itself.

These are totals, not rates: the data only covers trains that were delayed or cancelled. Through trains that call at Thatcham aren’t included in the cancellation figures. The most recent periods aren’t final, and causes can change if they’re disputed.

Compared with other operators

ORR’s figures for every operator, for the same months as the Thatcham data: station stops reached less than 5 minutes late, with cancellations counted as failures.

Results

• GWR ranks 16th of 23 operators (77%), well below the Great Britain figure (85%).
• The Elizabeth line, on the same tracks between Reading and Paddington, manages 89%. The best operators (Merseyrail, ScotRail) are above 92%.
• Within 15 minutes the gaps narrow - GWR 90%, Great Britain 93%, Elizabeth line 94% - so GWR’s weakness is mostly in the smaller delays and cancellations.
• For comparison, Thatcham’s own journeys - measured as whole journeys - arrived within 5 minutes about 65% to 70% of the time to and from London, and 75% to 88% to and from Reading and Newbury.

ORR measures arrivals at every station stop, and the cancellation adjustment is a close approximation rather than a published figure. Sources: ORR passenger rail performance statistics.

The official measures

Trains calling at Thatcham against Great Western Railway and the national average, per four-week rail period, using the same measures as the official statistics.

Why these look better than the journey figures: they count trains at Thatcham, not journeys. A train that leaves Thatcham on time and ends its journey at Reading counts as on time, and only cancellations at Thatcham itself count - not trains that end early, skip the stop, or are cut back or removed from the timetable in advance. Thatcham is measured on departure and the national figures on arrival, which works slightly against Thatcham.

Conclusion: the service isn’t broken - most trains run, and most run close to time - but the official measures miss a lot of what passengers experience.

Thatcham’s data starts on 8 July 2026, part-way through period 4 of 2026-27. National and operator figures are published 3 to 4 weeks after each period ends.

Lateness

A cancelled train counts as late in all of these charts. ORR leaves cancelled trains out of its punctuality figures, so the national and operator lines are ORR’s figures scaled down by their own cancellation rate, which is a close approximation rather than a published number.

Share of scheduled stops reached early or less than a minute late. Thatcham is measured when the train departs and the national and operator figures when it arrives, so this comparison is harsher on Thatcham than on the other two lines.

Share of scheduled stops reached early or less than three minutes late, on the same basis.

Average minutes late, with a cancelled train costing 1.5 times the gap between trains, the rule ORR uses for average passenger lateness. At Thatcham that is 40 minutes (90 on Sundays), close to the 41 minutes people actually waited on average for the next train. The national and operator figures are ORR’s average passenger lateness, which also weights by passenger numbers and is measured at main stations. Thatcham has no passenger numbers, so its line is an average per train.

Cancellations

For Thatcham, the share of scheduled calls at the station that were cancelled. The national and operator figures are ORR’s Cancellations measure, which counts whole trains, with a part-cancelled train scoring a half.

The data

Notes on the Thatcham figures: rail replacement buses are excluded. Period 4 of 2026-27 covers 8 to 25 July only, the latest period is still in progress, and no data was collected for 27 and 28 August or 28 September 2026.

Sources: Realtime Trains for Thatcham; Office of Rail and Road passenger rail performance statistics for Great Western Railway and Great Britain.