Fire Seasons Without End: The Evidence on Adapting to Europe’s Wildfire Risk

1Background: a structural, not seasonal, risk

The European Forest Fire Information System (EFFIS) recorded 1,079,538 hectares burned across 25 of the 27 EU member states in 2025, from 7,783 fires. It is the highest burned area in the EU since EFFIS began in 2006, nearly double the 2006–2024 average. Germany, Spain, Cyprus and Slovakia hit all-time records.^[1] A prolonged heatwave in the first three weeks of August triggered 22 very large fires in Portugal and Spain almost simultaneously, which burned 460,585 hectares, or 43% of the EU total.^[1] About 39% of the burned area lay within the Natura 2000 network of protected sites.^[1]

The 2025 season also showed a pattern. The JRC identifies an earlier start, more frequent and intense heatwaves, and fires reaching higher latitudes than historically observed.^[1] The European Commission describes the resulting risk as no longer seasonal, but structural.^[8]

The 2026 season has also been severe, though below 2025 on the cumulative measure. As of 23 September 2026, EFFIS reported 678,978 hectares burned in the EU, compared with 1,026,104 hectares by the same date in 2025 and a 2006–2025 average of 355,397 hectares (Figure 1).^[2] This brief asks three questions. What drives the risk? What does the evidence say about prevention and preparedness? And how well are people protected when fire reaches communities?

2What drives the risk

Weather. A study by World Weather Attribution (WWA) of the August 2025 fires in Spain and Portugal analyzed fire-conducive weather using the Daily Severity Rating. In today’s climate, which has warmed by 1.3°C since preindustrial times, the extreme conditions that drove the fires are expected about once every 15 years. Compared with a 1.3°C cooler climate, that is an increase in likelihood of about a factor of 40 and an increase in intensity of about 30%. The event would only be expected less than once per 500 years without climate change. The fires coincided with a 16-day heatwave in Spain, the most intense on record.^[4]

The 2026 analysis found a similar signal. Conditions with a seven-day Daily Severity Rating as extreme as those in central Spain in July 2026 are now at least 20 times more likely than in a pre-industrial climate, and are expected about once every 6 years. In southwest France the event is expected about once every 20 years, with an increase in likelihood of at least a factor of two that WWA calls conservative.^[3]

Weather whiplash. WWA links the 2026 fires to a sequence of above-average winter rainfall, which promoted vegetation growth, followed by anomalously dry and warm conditions that dried the fuel. In Spain, January 2026 was the wettest January of the last quarter century, and rainfall in the area affected by the Ávila–Madrid–Toledo fire was 175% higher than usual.^[3]

Fuel and landscape. Rural depopulation and an ageing population across parts of Portugal and Spain have left forest land unmanaged, creating dense fuel loads, and with fewer people and less traditional grazing, natural vegetation control has sharply declined.^[4] The Gironde fire burned through a region dominated by more than 800,000 hectares of intensively managed maritime pine monocultures, whose needle litter, ladder fuels and connected canopies allowed very rapid spread and crown fire.^[3] A review of Mediterranean-type regions lists agricultural land abandonment, expansion of settlements into fire-prone areas, and large, poorly managed plantations of highly flammable species among the land-use changes that raise fire hazard.^[5]

The same review notes that, despite extraordinary expenditure on suppression, most inter-annual variability in burned area in these regions is still explained by fire weather.^[5]

3Prevention works, up to a point

The most important finding for planning is that prevention has measurably worked, and that it can still be overwhelmed. WWA reports that Spain’s average annual burned area was around 232,000 hectares from 1975 to 1994, and fell to around 80,000 hectares in 2006 to 2024, mainly thanks to improved prevention measures and harsher sentences for those responsible for starting fires. Yet in 2025, one region burned to a degree not seen for three decades, with an area four times the annual average of the previous two decades.^[4] Portugal, proportionally the most affected country in Europe, has a mean annual burned area of around 115,000 hectares since 1980 and an all-time record of 540,000 hectares in 2017.^[4]

Moreira and 21 co-authors argue that this pattern is a trap. Policies that allocate most investment to suppression contribute to ongoing fuel accumulation and landscape-level fuel continuity, which then precludes suppression under extreme fire weather and produces more severe and larger fires. They recommend that policy and spending be rebalanced between suppression and mitigation of the negative impacts of fire, and that success be measured by avoided socio-ecological damage, not by area burned.^[5]

The European Commission’s wildfire strategy, adopted on 25 March 2026, moves in this direction. It covers prevention, preparedness, response and recovery, promotes fire-resilient landscapes through sustainable land management and nature restoration, and strengthens early warning and monitoring through EFFIS and Copernicus, while also boosting firefighting capacity.^[8] WWA’s 2026 analysis reaches a similar conclusion, saying reducing risk will require not only effective suppression and cross-border coordination but greater emphasis on prevention through risk-sensitive landscape planning and management and reducing exposure.^[3]

Prescribed burning is the tool most often cited. A review of its use in southern Europe finds neutral or positive effects on soils and biodiversity, but concludes that the limited extent of current applications is unlikely to reduce wildfire hazard, and that adoption has been slow, uneven and inconsistent.^[9] Moreira et al. describe it as a very cost-effective fuel treatment, hindered by bureaucracy, cultural resistance, perceived risk and resource availability.^[5]

4Protecting people when fire arrives

The 2018 fire at Mati, Greece, killed 104 people and is described as the second-deadliest wildfire of the 21st century.^[6,7] A model study notes that no warnings were issued by authorities during the event, so people acted on environmental cues.^[6] Using an agent-based model of the fire, the authors found that multimodal travel, staged evacuation and increased shelter capacity lead to a more successful evacuation.^[6] The results are simulations, but they point to design choices that emergency managers control.

Preparedness has not caught up with that lesson. A survey of Greek municipalities for the 2021 and 2022 fire seasons found that 82.1% did not conduct evacuation exercises, 64.1% did not provide evacuation training for residents of wildland–urban interface areas, and almost half lacked tailored evacuation plans for vulnerable communities. Civil protection representatives often described their municipalities as very well prepared, but the study found a misalignment between perceived and actual preparedness, and it recommends alternatives such as shelter-in-place where evacuation is impractical.^[7]

The 2026 fires showed the scale at which evacuation now operates. WWA reports that the Gironde fire burned about 40,000 hectares, the second-largest area devastated by a single wildfire in France since the Second World War, destroyed at least 240 homes and led to around 220,000 residents and tourists being evacuated. It generated a pyrocumulonimbus, or fire-driven thunderstorm. Particulate matter in Bordeaux exceeded the level considered fair by a factor of 20, and the national strategic reserve of FFP2 masks was released to pharmacies. In Spain, more than 80,000 people were evacuated and nearly 100,000 were advised to stay indoors because of poor air quality.^[3]

Fires are also beginning to strain the response system itself. The EU Civil Protection Mechanism was activated 17 times by 15 August 2025, more than the total for the whole of the 2024 season, with requests from six countries within an eight-day period.^[4]

5Implications for disaster risk management

Analysis. This section is Disaster Science’s interpretation of the evidence above. It is not drawn from the cited sources.

Measure what you want to protect. Burned area is the headline statistic, but it hides lives, homes and health. Agencies should publish avoided-harm indicators alongside area, including evacuations, structures lost, and smoke exposure.

Plan for two kinds of year. Prevention appears to cut average losses, but the extreme years drive harm. Preparedness budgets, mutual aid and evacuation plans should be sized for the weather that overwhelms normal capacity, not for the average season.

Treat evacuation as an engineered system. Routes, staging, shelters, transport for people without cars, and clear triggers are design variables. Regular exercises are the cheapest evidence of whether a plan works, and the Greek survey suggests most municipalities are not yet running them.

Match fuel policy to livelihoods. The fuel problem follows from the loss of grazing and farming. Prevention programs that pay people to manage land, rather than only prescribing what land should look like, address the cause as well as the symptom.

Do not forget smoke. The Bordeaux readings and the advice to stay indoors in Spain show that health protection from smoke is part of wildfire risk management, with its own supplies and communication plans.

6Uncertainty and limitations

Rapid attribution. WWA describes its analyses as super rapid, using observations only, and the 2026 study was written on 30 July while fires were still burning. For France, the authors say the statistical model fits less closely and the factor of two is a conservative estimate.^[4,3]

Counts change. Burned area, evacuations and homes destroyed are early figures and are refined as satellite imagery and official tallies arrive. The 2026 totals are year-to-date as of 23 September and the season is not over.^[2,3]

Evidence gaps. This brief found no outcome evaluations for structure hardening or wildland–urban interface planning in Europe, and the evacuation findings combine a simulation with a municipal survey. Portugal’s post-2017 reforms are not assessed here. The review by Moreira et al. is an expert perspective, not a meta-analysis.^[5,6,7]

Scope. This brief synthesizes official reports, rapid attribution studies and peer-reviewed papers. It presents no original field data and is an editorial synthesis that has not undergone peer review.