Alterations in the hydrological cycle following wildfire due to the loss of ground cover vegetation and changes in soil properties have been documented in many studies. Nevertheless, the rapid process of vegetation recovery reduces such negative effects. Vegetation cover before fire, fire severity, and geophysical properties are important factors that control spatial discontinuities involved in the vegetation-covering process. The objective of this study was to estimate the probability of high erosion in order to map erosion-sensitive areas after fire. The analysis was carried out in different plant communities burnt by summer wildfires in the pre-Pyrenean area (Spain). Three-year Landsat Thematic Mapper (TM) images have been used for mapping wildfire areas and severity levels. Conversion to spectral reflectance has been applied for radiometric correction by normalizing topographic and atmospheric effects. Likewise, other physical variables have also been incorporated into the geographic information system (GIS): vegetation types, parent material, illumination, slope, aspect, and precipitation. The dependent variable has been characterized by means of fieldwork and a photointerpretation process based on high-resolution digital aerial orthophotographs taken 11--12 years after the fire. Different logistic regression models have been used for mapping the probability of erosion. Results indicate that prefire normalized difference vegetation index values and aspect are the most important variables for estimating erosion-sensitive areas after fire (Nagelkerke r2 = 0.66; Kappa values = 0.65). Finally, the use of nonparametric models with environmental digital information based on GIS can facilitate the management of burnt areas.