← back to the reports
DR-002 Passenger express · Renfe, Spain 2013

Santiago de Compostela — An Express at 180 on an 80 km/h Curve

Killed
79
Railway
Renfe
Service
Alvia passenger express
Status
Driver

Summary

On 24 July 2013, at about 20:41 local time, a Renfe Alvia express on the Madrid–Ferrol service derailed on the A Grandeira curve at Angrois, roughly four kilometres short of Santiago de Compostela station in Galicia, north-western Spain. The train entered the 80 km/h curve at about 180 km/h; all thirteen vehicles left the rails and several overturned, and 79 people were killed and some 140 injured. It is the deadliest rail accident in modern Spanish history. (The toll was briefly reported as 80 because of a misidentification of remains; the confirmed figure is 79.)

The train was an Alvia Class 730, a variable-gauge set able to run on both high-speed and conventional lines. For most of the route from Madrid it had run on high-speed track equipped with the European train-control system ERTMS/ETCS, which can automatically enforce speed limits. As the line approached Santiago it transitioned to conventional track fitted with the older ASFA system, which warns the driver of restrictions but cannot itself slow or stop an overspeeding train. The A Grandeira curve sat just past that transition, where a long stretch of 200 km/h running gave way abruptly to an 80 km/h limit — and where the responsibility for braking in time rested entirely on the driver. In the moments before the curve, the driver, Francisco José Garzón Amo, was on the train's telephone with a Renfe controller discussing the route ahead and consulting a document; he braked late and hard, but the train was still doing about 179 km/h when it derailed.

The official investigation was conducted by the Comisión de Investigación de Accidentes Ferroviarios (CIAF), the rail accident investigation commission within Spain's Ministry of Public Works (Ministerio de Fomento). Its report attributed the accident to human error — the driver's failure to reduce speed for the curve, distracted by the phone call and his loss of position — and treated this as the only contributing cause. That conclusion drew sustained criticism. In 2016 the European Commission asked the European Union Agency for Railways (the ERA, successor to the European Railway Agency) to review the inquiry; the agency questioned CIAF's independence — the team had included staff from the infrastructure manager ADIF, the operator Renfe, and the consultancy Ineco — and faulted the report for emphasising the single human error while underweighting the systemic and signalling factors: the absence of automatic train protection at the curve, the abrupt speed transition, and the unaddressed derailment risk that contractors had earlier flagged.

The criminal proceedings reflected that wider picture. After a trial that began in 2022, a court in 2024 convicted both the driver and a senior ADIF safety official, finding that the failure to assess and mitigate the curve's risk, as well as the driver's conduct, contributed to the deaths.

Timeline

24 July 2013, afternoon
Departure
Renfe Alvia service 04155, an Alvia Class 730, leaves Madrid bound for Ferrol in Galicia, running first on high-speed track under ERTMS/ETCS control.
En route
Transition to conventional track
As the line nears Santiago de Compostela it changes to conventional track equipped with the older ASFA system, which warns but cannot enforce a speed reduction.
24 July, ~20:39
Approach at speed
The train approaches the A Grandeira curve at Angrois after a stretch of high-speed running; the curve is limited to 80 km/h.
24 July, ~20:40
The phone call
The driver is on the train telephone with a Renfe controller discussing the approach to Ferrol and consulting a document, his attention divided in the seconds before the curve.
24 July, ~20:41
Late braking, then derailment
Roughly 250 m before the curve the train is doing about 195 km/h; the driver brakes hard but the train enters and derails at about 179–180 km/h. All thirteen vehicles leave the rails; several overturn.
24 July, evening
The toll
Seventy-nine people are killed and around 140 injured; the figure is briefly reported as 80 owing to a misidentification of remains.
July 2013 onward
Driver charged
The driver, Francisco José Garzón Amo, is detained and later charged with multiple counts of professional recklessness causing death and injury.
2014
CIAF report
Spain's rail accident investigation commission, within the Ministry of Public Works, publishes its report attributing the derailment to driver human error as the sole contributing cause.
2016
EU scrutiny
The European Commission asks the EU Agency for Railways to review the inquiry; the agency criticises CIAF's independence and its failure to weight systemic and signalling factors.
October 2022
Trial opens
The driver and a former ADIF safety director go on trial in Santiago de Compostela.
26 July 2024
Convictions
The court convicts both the driver and the ADIF safety official, imposing prison terms, professional bans, and substantial compensation to the injured.

The Line and the Curve

The route from Madrid to the Galician coast was a hybrid by design. New high-speed infrastructure carried trains at 200 km/h or more for long stretches, and on it the European train-control system ERTMS, in its ETCS form, supervised the train continuously: had it approached a restriction too fast, the system could have braked automatically. The Alvia Class 730 was built to bridge worlds — a variable-gauge, dual-voltage set that could leave the high-speed network and continue onto the older conventional lines that still reach much of Spain.

The handover between those worlds was where the danger concentrated. As the train neared Santiago, the line reverted to conventional track equipped only with ASFA, an older system that warns the driver and can apply the brakes for certain signal violations, but does not enforce a continuous speed profile through a curve. The A Grandeira curve at Angrois lay just beyond that boundary, where an 80 km/h limit followed almost immediately on a long, fast approach — one of the sharpest reductions on the route — and nothing on the conventional side would physically stop a train from carrying its high-line speed into it. The on-board ETCS that might have caught such an error had been disconnected in June 2012, owing to operational problems, leaving the curve protected only by the driver's judgement and ASFA's warnings. Contractors who built the line had, the later court proceedings established, identified the curve as a derailment risk and suggested an ERTMS overlay, but no formal risk assessment had been carried out by the infrastructure manager.

This is the systemic frame the CIAF report was later criticised for underplaying: geometry, speed transition, protection system, and disconnected equipment together meant that a single lapse of attention at one point on the line had no automatic safety net beneath it.

The Eleven Seconds

For most of the run the train behaved exactly as it should. The recorded data show it approaching Angrois at roughly 195 km/h some 250 metres before the curve — appropriate for the high-line speed it had been holding, but far too fast for the 80 km/h limit ahead. In the seconds that mattered, the driver was on the train's internal telephone with a Renfe controller, discussing the approach to Ferrol and, the investigators found, consulting a document. His attention was divided, and by his own later account he lost track of exactly where he was on a line whose abrupt change of character left little margin for such a lapse.

He braked hard when he recognised the curve, but it was too late to shed enough speed. The train entered A Grandeira at about 179 to 180 km/h — well over twice the limit. The lateral forces overwhelmed it; all thirteen vehicles derailed and several overturned, some striking a retaining wall and a service road beside the line. The crash was captured on a trackside camera, fixing the speed and dynamics beyond dispute. Seventy-nine people died and roughly 140 were injured, making it the worst Spanish rail disaster in decades.

The mechanism is, in the cold terms of the case file, an overspeed derailment caused by the driver's failure to reduce speed for a known curve while distracted — the reason the Finding is recorded as Driver. But the record also shows that nothing in the line's equipment was positioned to save him from that error, and that is the part the official inquiry was faulted for setting aside.

The Verdict and the Quarrel Over It

The CIAF, sitting within the Ministry of Public Works, conducted the official technical investigation and concluded that the cause was human error: the driver had not complied with the speed and route information in his timetable and route plan, distracted by the phone call moments before the curve, and the commission treated this as the only contributing cause. As a finding about what physically caused the train to leave the rails, it was not in doubt — the speed data settled that.

What was contested was the scope of the inquiry. In 2016 the European Commission asked the EU Agency for Railways to review CIAF's work, and the agency's assessment was pointed. It found that CIAF had not satisfied the European requirement that an accident investigation body be independent of the infrastructure manager, the railway undertaking, and any party whose interests might conflict with the investigation — singling out the presence of ADIF, Renfe, and Ineco staff on the team. It judged that the report's emphasis on the single human error came at the expense of the underlying and root causes, most likely to include the organisational actions of ADIF and Renfe: the design of the line, the train, and the signalling, and the absence of automatic train protection where a steep speed reduction met conventional track. The agency did not dispute that the driver sped into the curve; it disputed an inquiry that stopped there.

The courts ultimately reached the broader conclusion the agency had pressed for. After a trial that opened in 2022, a verdict in July 2024 convicted not only the driver, for failing to slow for the curve, but also a former ADIF director of safety, for the failure to assess and guard against its known derailment risk; prison terms, professional disqualification, and large compensation orders followed. The case stands as a caution against confining a verdict to the last person in the chain when the safety system around that person had been left incomplete.

The Five Factors

01
Driver attention at the one point that demanded it
The driver was on the telephone and consulting a document in the seconds before a curve that required a sharp speed reduction, and he braked too late. Where a single moment of divided attention has catastrophic consequences, attention is itself a safety-critical resource that procedures must protect — not a thing to be spent on routine calls at the most demanding point on the line.
02
An abrupt speed transition with no automatic enforcement
A long high-speed approach gave way to an 80 km/h curve on conventional track protected only by ASFA, which warns but cannot brake. Designing a route so that an enormous speed reduction depends entirely on human compliance, with no continuous train protection, places the whole margin of safety on one person at the worst possible spot.
03
Disconnected on-board protection
The ETCS equipment that could have enforced the limit had been switched off in 2012 for operational reasons. A protective system that is disabled to keep trains running smoothly is no protection at all; removing an automatic safeguard demands a documented compensating measure, not silence.
04
A flagged risk left unassessed
Contractors had identified the curve as a derailment hazard and suggested an ERTMS overlay, but no formal risk assessment was done. A known hazard that is noted and then not analysed or mitigated is an open invitation; risk identification is worthless without the assessment and action that are supposed to follow it.
05
An investigation that stopped at the driver
The official inquiry treated the driver's error as the sole cause and was criticised by the EU agency for its lack of independence and its neglect of organisational and signalling factors. An investigation that names only the last human in the chain forecloses the systemic fixes — protection, geometry, risk process — that prevent the next event.

Aftermath

The Santiago derailment became a long, bitter case in Spain — for the bereaved, who fought for years to widen the inquiry beyond the driver, and for the European institutions that questioned how Spain investigates its own accidents. The EU agency's 2016 critique put national independence of accident investigation under scrutiny across the bloc and pressed Spain toward reforming how CIAF was constituted. In the years after, attention to automatic train protection at junctions between high-speed and conventional lines sharpened, and the disconnection of on-board ETCS without compensating safeguards became a cautionary example in European rail-safety practice.

The criminal outcome, reached in 2024, was notable for refusing to confine blame to the man in the cab. By convicting both the driver and a senior infrastructure-manager official, the court endorsed the view that the catastrophe was the product of a driver's error and an organisation's failure to protect against the very curve that error met. For the families, it was a partial vindication of a decade's insistence that the question was not only why the driver sped, but why nothing on the line was there to stop him.

Lessons

  1. Never let an enormous speed reduction depend on human compliance alone; a curve where high-speed track meets a low limit needs continuous automatic train protection, not just a warning system.
  2. Treat the disabling of an on-board protection system as a safety event in itself, requiring a documented, equivalent compensating measure before trains keep running.
  3. Act on flagged hazards: a curve identified as a derailment risk must trigger a formal risk assessment and mitigation, not a note that is filed and forgotten.
  4. Protect driver attention at the most demanding points of a route; procedures should not place routine communication tasks where a moment's distraction is fatal.
  5. Investigate to the root: an inquiry that stops at the last human error, and is not independent of the operator and infrastructure manager, leaves the systemic causes unfixed for the next train.

References