On 19 June, two trains collided at Elstow, south of Bedford which sadly resulted in the death of a train driver and many passengers suffering major injuries. Although the UK has one of the safest railways in Europe, this shows that there is always the potential for a serious accident. Three days after the accident, the Railway Accident Investigation Branch (RAIB) released its statement about the accident. This showed that Train 1H46, which was formed of a four-car 2003-built Class 360 unit, passed a red signal and ran into the back of Train 1B67, a brand new five-car Class 810, at a speed of around 79 km/h.
Historic reports into such accidents would typically conclude that ‘the driver was driving without due attention to the signals.’ Yet it is wrong to conclude that the driver was at fault because a red signal was passed as this will be the result of various factors which will not be known until RAIB’s report into this incident is produced.
This will identify the causal factors which resulted in the immediate cause and any underlying factors associated with the organisations concerned. It will also make recommendations to prevent a recurrence which will relate to the identified causal and underlying factors.
Significant risk reduction
To understand the risk profile and current level of risk from train accidents, it is instructive to consider past accidents. Between 1980 and 2026 there were 34 fatal passenger train accidents as shown in Table 1. Network Rail has been in existence since October 2002 which is almost exactly half of this 46-year period, therefore a direct comparison can be made between the number of accidents between 1980 and 2002 when British Rail and Railtrack were responsible for rail infrastructure and the Network Rail era from 2003 to 2026.
This shows that the number of fatal passenger train accidents and fatalities from such accidents during the Network Rail era was respectively 17% and 8% those of in the British Rail and Railtrack period. Thus, in the Network Rail era there were fewer fatalities per fatal train accident. This is due to improvements in train crashworthiness as discussed later in this feature.
Much of this reduction in fatal accidents is due to the dramatic reduction in SPADs. Indeed, prior to the Elstow accident, the last fatal SPAD accident was the 1999 Ladbroke Grove tragedy which was 27 years ago. Table 1 shows that between 1980 and 2002 there were 11 fatal SPAD and Buffer stop collision accidents, which was about one every two years.
This large reduction in the number of SPADs was due to a combination of driver training emphasising what is now termed professional driving and the Train Protection and Warning System (TPWS) programme as reported in issue 210 (Sep-Oct 2024). TPWS was mandated by the 1999 Railway Safety Regulations which required its fitment at all signals protecting a passenger line junction and buffer stops and to speed reductions where speed was reduced by 30% or more from an initial speed exceeding 60mph. When the programme was completed in 2003, more than 12,000 signals had been fitted with TPWS which was around 30% of all signals. Ongoing signal risk assessments and some resignalling projects have increased the percentage of signals now fitted to around 40%. The signal passed at danger at Elstow was not fitted with TPWS.
Although TPWS is not a failsafe system and is not fitted at all signals, it has demonstrably resulted in a significant improvement in railway safety. Yet, as the Elstow accident shows, SPADs still present a risk. Hence it is likely that TPWS fitment criteria will be considered by RAIB’s investigation into this incident.
Changing risk profile
The risks of passenger train operations have changed over time due to action taken by the industry and changing external factors. To examine this with a larger data set than fatal train accidents, the immediate causes of 69 passenger train accidents since 1980 that resulted in major injuries were analysed to produce Table 2 (right). This shows the increase and decrease of various types of accidents before and after the creation of Network Rail.
The types of incidents, together with their percentage of all accidents, for which there have been an increase since 2002 are: earthworks (15%); poor adhesion (10%); collisions during signalled moves (10%); and level crossings (25%). The impact of the changing climate on +150-year-old earthworks presents significant challenges. Climate change may also be a factor in the increase of low adhesion incidents as milder, wetter autumns and extended growing seasons increase rail corridor vegetation.
Various factors have increased level crossing risk including societal changes affecting how user worked crossings are used. Rail Engineer has published various features explaining the risks associated with level crossings and the action being taken to reduce this risk, with a recent report in Issue 219 (Mar-Apr 2026). Though UK level crossing incidents are well-below the European average, there were 67 train collisions with vehicles on level crossings in the 10 years to 2023.
The types of accidents that have been significantly reduced are SPADs and buffer stop collisions as previously discussed, broken rails, train defects, and signaller error. Significant improvements in rail management, as described in Issue 202 (May-June 2023), reduced the number of broken rails from 952 in 2000 to 70 in 2023/24. The procurement of modern trains will have reduced the potential for train defects. The potential for unsafe signaller errors has been reduced due to the closure of an estimated 3,000 mechanical signalboxes since 1980. As a result, under 100 remain as Network Rail aspires to eventually control the entire network from 12 Regional Operating Centres.
Crashworthiness
The RAIB investigation will consider the crashworthiness performance of both trains during this 79km/h collision. In this respect it is instructive to consider how rolling stock behaved in previous rear-end collisions. As such collisions are quite rare, this requires looking back to the 1970s to consider the fatal accidents at Shields Junction in 1973, Hassocks in 1978, and Clapham in 1988. These had collision speeds of respectively 82 km/h, 76 km/h and 56 km/h.
In the Shields Junction crash, a three-car Class 303 EMU travelling at 104km/hr hit a train formed of 2 x three-car Class 126 DMUs which was moving at 22km/hr (i.e a closing speed of 82km/h). In this collision, the leading coach of the EMU was forced underneath the DMU which was forced upwards into the overhead line equipment. This crushed the cab and much of the passenger compartment of the EMU’s lead coach to solebar level. The body structure of the trailing coach of the DMU was also crushed. However, other than the DMU’s trailing coach, there was no derailment and the adjacent lines were not fouled.
In the Hassocks crash, a 12-car EMU ran into the back of an eight-car EMU at an estimated 76km/hr. Both trains were formed on four-car Class 420 and 421 units. On impact the body of the last coach of the front train was demolished as it deflected to foul the adjacent line and its next coach was thrown up against a bridge. The body of the first coach of the rear train was demolished as it passed under the last three coaches of the front train.
At Clapham Junction, one train ran into the back of another train at 56 km/hr. This threw its rear coach above a three-metre-high concrete wall while the front coach of the rear train veered into the path of an oncoming train. The high death toll in this crash was due to the passenger compartments of the Mark 1 coaches that hit other coaches being breached due to the lack structural strength. This was because, unlike the monocoque construction of the later Mark 2 and 3 coaches, the main structural member of a Mark 1 coach is its underframe.
The fatalities and injuries from these three accidents were largely the result of passenger compartments being breached, the lack of overrun protection and, in the case of Clapham, coaches forced onto an adjacent line. The damage to the trains involved in the Shields Junction and Hassocks accidents indicates the energy involved in rear-end collisions at around 80 km/hr.
Although this was about the same speed as the Bedford crash, it would seem that no passenger compartments were breached and both trains remained upright on the line. Although one bogie was derailed, this did not block the adjacent line but might have fouled it. Hence, in respect of crashworthiness, the trains involved in the Bedford accident performed well compared with these earlier crashes
This is also indicated by the crashworthiness standard EN15227 which recognises that there are limits to what any structure can withstand. It requires main line trains to withstand a collision with another train at the low speed of 36 km/h. At 78 km/hr, the Bedford collision had four times the energy of this limit.
Yet, RAIB may well find that that there are crashworthiness lessons from this accident particularly in respect of coach interiors. As far as is reasonably practical, the effects on those inside a coach of a train brought to a dead stop in a rear-end collision have to be considered.
Maintaining safety
Throughout railway history there have been continuous safety improvements driven by learning from accidents. As a result, UK rail travel is very safe and fatal accidents such as Esltow are rare. But this accident is a stark reminder of the ever present risks associated with railway operations, and of the consequent need for constant learning and vigilance.
It is also important that those who have to implement and monitor safety related processes understand the need for them. Accidents are now so rare that lessons from accidents in the distant past might be forgotten. In 1989, the multi-fatality Clapham Junction accident was a huge shock to the industry, particularly the S&T community. Yet that was almost 40 years ago and there were recently similar wrong side signalling failure incidents at Cardiff (2016), Waterloo (2017), Dalwhinnie (2021), and Wingfield (2022).
A lack of learning from a previous incident was shown in the RAIB report into overspeeding at Peterborough’s Spital Junction in 2023 which noted that lessons had not been learnt from a similar previous incident in 2022. In its 2025 annual report RAIB identified such overspeeding incidents, together with asset integrity and level crossings, as themes from its investigations that put trains at risk.
This report also notes that an absence of high consequence accidents is not necessarily an indication of safety as the difference between a near miss and a serious accident can be a matter of luck. Hence, while the industry’s good safety record is a credit to all concerned, there is no room for complacency.
Article lead image photo credit: Rail Accident Investigation Branch