Background <p>Wildfires have long shaped the ecosystems of southwestern Europe, but in recent decades, their frequency, size, and severity have become a widespread socio-ecological threat. In this study, we assessed the 2025 extreme fire season in the north western Iberian Peninsula through remote sensing and modelling workflow aimed at identifying drivers of burned area and fire severity. Large wildfires (≥ 500&#xa0;ha) occurred during the summer fire season (June–September) were extracted from the EFFIS database, while Sentinel-2 imagery was used to estimate spectral fire severity. Pre-fire vegetation, topography, and fire weather were integrated as explanatory variables by fitting separate Random Forest (RF) regression models.</p> Results <p>Our results revealed that a total of 66 large wildfires burned over 520,000&#xa0;ha across Spain and Portugal, with only one-quarter of these events accounted for over 75% of the total burned area. Grasslands, shrublands, and transitional shrub-woodlands were the most affected land cover types, with protected areas disproportionately affected mainly in NW Spain where 40% of burned area was located in Natura 2000 sites. Fire severity patterns were also remarkable, with high and moderate-high severity classes exceeding 65% of the total burned extent. RF models revealed that this anomalous burned area was driven by top-down atmospheric drivers, particularly exceptionally high vapor pressure deficit (VPD &gt; 4&#xa0;kPa), strong winds and drought conditions during prolonged heatwave periods. Conversely, fire severity was mainly associated with bottom-up drivers, namely high fuel loads, extensive forest cover, and complex topography.</p> Conclusions <p>Our findings indicate the need to strengthen adaptive management strategies into conservation and land-use policies in order to reduce fire hazard and associated ecological impacts under worsening climatic conditions.</p>