Guide
The Iberian blackout of 28 April 2025: what happened, second by second
Continental Spain and Portugal lost power in about twenty seconds. The official investigations took eleven months, and the question of who pays is still open.
On this page
The first warning came at 12:03 on a sunny Monday. For about five minutes, power flows across southern Europe began to swing in a slow oscillation, and control rooms in Madrid and Lisbon watched it. A second, wider oscillation followed at 12:19. At 12:33 CEST on 28 April 2025, continental Spain and Portugal lost their entire electricity supply, about 31 GW of demand disconnected within seconds, and the Iberian Peninsula went dark from Galicia to the Algarve.
According to ENTSO-E it was the first time that a cascade of generator disconnections driven by rising voltage had led to a blackout in the Continental European synchronous area. Most blackouts are about too little power. This one was about voltage that climbed too high.
Some physics helps here. On an AC grid, voltage at each substation is held within a narrow band by trading reactive power: generators and devices either inject it to push voltage up or absorb it to pull voltage down. A long high-voltage line carrying little current behaves like a capacitor and pushes voltage up on its own, so a grid that is exporting at noon on half-empty lines needs plenty of machines ready to absorb. Large synchronous generators do that automatically through their excitation systems; inverter-based solar and wind plants can do it too, but only if their controls are set to, and in Spain in April 2025 many were not. The terms are explained in the glossary.
What the grid looked like at 12:30
The day was ordinary for spring on the peninsula: strong sun, moderate demand and cheap power flowing out. According to the expert panel’s final report, cheap solar power had pushed Spain’s exports to 5 GW in total during the morning; by 12:24 the physical flow to France was 866 MW against a scheduled 1,000 MW, and four minutes later it was down to 190 MW. Solar plants supplied the majority of Spanish generation. Synchronous machines, the gas, nuclear and hydro units that naturally stiffen voltage, were fewer than on a winter evening.
Spain’s operator Red Eléctrica had scheduled ten thermal plants to provide dynamic voltage control that day. According to the Spanish government’s report, one of them declared at 20:00 the previous evening that it would not be available, and it was not replaced for the central hours of 28 April.
The timeline, minute by minute
| Time (CEST) | Event |
|---|---|
| 12:03–12:08 | First oscillation, mainly within Spain and Portugal |
| 12:19–12:22 | Second oscillation, an inter-area swing across the continental grid |
| After 12:22 | Operators reduce exports to France and couple more lines in southern Spain to damp the swings; voltage rises |
| 12:32:00–12:32:57 | Output of large renewable plants falls by about 500 MW; about 208 MW of smaller wind and solar units drop out and net load in the distribution grids rises by about 317 MW |
| 12:32:57–12:33:18 | Plants trip at Granada (355 MW) and Badajoz (727 MW), then 928 MW more in Segovia, Huelva, Badajoz, Seville and Cáceres; with the earlier losses more than 2.5 GW is gone and voltage climbs further |
| 12:33:21–12:33:24 | AC lines and then the HVDC link to France disconnect; the Iberian system collapses |
| Afternoon to early 29 April | Restoration from France, Morocco and Portuguese black-start units |
The damping measures are the uncomfortable part of the story. Reducing exports and switching more lines into service are standard answers to an oscillation. Both also raise voltage on a lightly loaded grid, and on 28 April they did exactly that, at a moment when too few plants were absorbing reactive power to bring it back down.
From there the sequence fed itself. As voltage rose, protection relays at generating plants tripped them off. Each lost plant removed some of the little voltage control left, so voltage rose again, tripping the next one. Frequency fell to 48 Hz, the last step of automatic load shedding, which could not prevent the collapse. Protection devices then opened the ties to France, which, the final report notes, stopped the disturbance spreading into the rest of the continental grid.
What the official reports found
Three investigations matter, and they do not carry the same weight.
The Spanish government (17 June 2025). A committee set up by the government delivered its non-confidential report in 49 days. Its conclusion was an overvoltage of “multifactorial” origin: a system with insufficient dynamic voltage control, persistent oscillations and disconnections of plants, some of which “occurred inappropriately” before reaching their authorised protection limits. It named the unreplaced thermal plant, found no evidence of a cyber-attack, and recommended stronger oversight of whether generators actually meet their obligations, voltage control from renewable plants, faster grid planning and more interconnection. The report is published on the government’s site.
Red Eléctrica (18 June 2025). The operator’s own report laid the weight on conventional plants. Its leadership said publicly that had those plants controlled voltage as required, there would have been no blackout.
The ENTSO-E expert panel (factual report 3 October 2025, final report 20 March 2026). This is the most complete account. The 49-member panel included transmission operators, regional coordination centres, ACER and national regulators, and was chaired by experts from two unaffected operators. The final report spreads responsibility across the system:
- several conventional generators met their reactive power reference in fewer than 75% of hourly samples, the minimum fulfilment ratio set by Spain’s Operational Procedure 7.4;
- many renewable plants were running in a fixed power factor mode, which gave them almost no ability to respond to voltage;
- some plants disconnected before voltage reached the thresholds in their connection agreements, with overvoltage protection set below regulatory requirements;
- shunt reactors, the devices that absorb excess reactive power, were switched manually and therefore slowly;
- the operator lacked a real-time view of the gap between the reactive power it needed and what generators were supplying.
The panel issued 21 new recommendations and amended two from its investigation into the 2024 South-East Europe incident. They cover five areas: voltage control and reactive power, oscillations, disconnection behaviour and protection settings, the system defence plan, and restoration. It also said plainly that even much higher inertia would not have prevented the loss of synchronism, which cuts against the argument that the answer was simply more spinning machines.
Final, disputed and still open
It helps to separate three kinds of finding.
Settled. The sequence of events and the timings are no longer contested. Nor is the mechanism: a voltage collapse in the upward direction, in which rising voltage caused cascading disconnections. Neither the Spanish nor the European investigation found a cyber-attack.
Disputed. Who carries the blame. Red Eléctrica and the large generators, represented by the association AELEC, have given opposing accounts to the Spanish parliament, the generators arguing that the operator failed to schedule enough synchronous capacity after voltage spikes on 22–24 April. The competition and energy regulator CNMC has opened more than 30 sanction proceedings against Red Eléctrica and around ten energy companies; by the first anniversary none had produced a final ruling, and the most serious breaches carry fines of up to €60 million.
Unknown. Some disconnections could not be explained. The final report records data gaps, especially for rooftop solar below 1 MW, where distribution operators did not have actual production data.
A fair reading of the reports is that the event was a failure of rules as much as of machines: voltage control had been treated as a job for a shrinking number of conventional plants, while the growing renewable fleet was neither allowed nor paid to do it.
France, INELFE and the Bay of Biscay
The peninsula’s thin connection to the rest of Europe shaped both the collapse and the recovery. Red Eléctrica describes Spain’s cross-border capacity with continental Europe as just 2%, far below the EU’s targets of 10% by 2025 and 15% by 2030, and the France border was congested in 67.6% of hours in 2024.
The main link is INELFE, two 1,000 MW HVDC circuits running underground through the eastern Pyrenees, 8.5 km of them in a tunnel, which doubled exchange capacity with France from 1,400 to 2,800 MW when it opened in 2015. On 28 April those circuits and the AC lines alongside them opened within three seconds of each other as the Iberian frequency collapsed. That separation is exactly what protection is for, and the HVDC link did its part by simply stopping; a stronger tie would not necessarily have saved Iberia, but a weak one leaves the peninsula with little outside help to lean on in the seconds that matter.
France then did the useful work. Ten minutes after the collapse, at 12:43, the first 400 kV line from France, Argia–Hernani, was back in service, and a second, Baixas–Vic, followed at 13:35. The tie to Morocco was energised at 13:04, while Portugal restarted from its own black-start plants.
The next step is the Bay of Biscay interconnection: two more 1,000 MW HVDC links with about 300 km of cable under the sea, raising exchange capacity from 2,800 to 5,000 MW. Construction is under way and the operators give 2028 for commissioning. Neither link fixes voltage control inside Spain, which is a domestic job; what they add is a larger neighbour to lean on, and fewer hours in which the border is full (the guide to grid congestion explains what a full border costs). Both sit in the directory of European power links alongside Europe’s other cross-border links.
Voltage control since the blackout
Two changes have followed on the ground, and you can watch both in Red Eléctrica’s own figures.
The first was defensive. Since May 2025 Red Eléctrica has run the system in a “reinforced operation” mode, scheduling more synchronous plants, mostly combined-cycle gas, throughout the day. By the end of April 2026 that had cost €711 million, about 2% of total system costs, according to the operator.
The second is structural. In June 2025 the CNMC approved a new version of Operating Procedure 7.4, a rulebook that had been requested in 2020 and that for the first time lets renewable plants be paid for dynamic voltage control. Red Eléctrica began authorising the first plants in October 2025 and the service took formal effect on 1 January 2026. By 30 April 2026, 14.5 GW of plant was providing voltage control by setpoint, 6 GW of it renewable.
The ENTSO-E recommendations now have to be written into national and European rules, and the CNMC’s sanction files, opened in April 2026, can take nine to eighteen months, so the first rulings are not due before early 2027. Whether a single company is ever found at fault for 28 April is not yet known.
Power links in this guide
Questions
What caused the Iberian blackout?
No single fault. The ENTSO-E expert panel's final report of 20 March 2026 describes a combination of oscillations, gaps in voltage and reactive power control, different voltage regulation practices, rapid output reductions and generator disconnections in Spain. Rising voltage tripped plants, which pushed voltage higher still, until the system lost synchronism at about 12:33 CEST.
Was the blackout caused by too much solar power?
Not in the way early commentary suggested. The reports point to how voltage was controlled, not to the share of solar as such. Many renewable plants ran in a fixed power factor mode that gave no voltage support, and some tripped earlier than their connection rules allowed, but conventional plants also delivered less reactive power than required. The final report says higher inertia alone would not have prevented the loss of synchronism.
Was it a cyber-attack?
The Spanish government's committee, reporting on 17 June 2025, found no evidence of a cyber-incident or cyber-attack as the cause. It did identify vulnerabilities, gaps and misconfigurations in the systems of some operators and recommended fixing them. The ENTSO-E panel's technical analysis also explains the sequence of events without any external attack.
How long did it take to restore power?
Restoration began in the afternoon of 28 April using the interconnections with France and Morocco and black-start units in Portugal. Portugal was back just after midnight, and Spain's transmission grid was back in full by the early hours of 29 April, around 16 hours after the collapse in the slowest areas.
Did the link with France help or hurt?
Both, at different moments. During the oscillations, operators cut exports to France, which the final report found contributed to the voltage rise. When the collapse came, the AC and HVDC links to France disconnected automatically as designed, which kept the rest of Europe stable. Afterwards France supplied the power used to restart Spain.