We model the occurrence of European wild rabbit in fragmented environments in a mountainous area of northwestern Spain (Gerês–Xurés Biosphere Reserve). We carried out a field survey by sampling the presence/absence of pellets in 237 plots (100 × 100 m) selected at random below an altitude of 800 m. For modelling purposes, we considered eight predictors related to vegetation, topography, human influence and heterogeneity. We obtained vegetation and ecological predictors from land use/land cover maps derived from Landsat Enhanced Thematic Mapper Plus images (acquired at the same time as the field data) and calculated vegetation indices by using a supervised classification method. We obtained topographical predictors from a Global Digital Elevation Model (GDEM) and used a generalized linear model to describe the occurrence of the European wild rabbit. The overall accuracy of the Landsat-derived map in Baixa Limia was 87.51 %, and the kappa coefficient was 0.85. The most parsimonious model included “grassland and crops”, “mean slope”, “distance to roads”, “urban settlements” and “ecotone scrubland-forest”. Five predictors were consequential, three of them with a positive sign for the presence of the species (scrub, urban settlements and ecotone scrubland-forest) and two with a negative sign (mean slope and distance to roads). The information on habitat requirements of European wild rabbit in the area provides a good framework for determining the habitat requirements of this keystone species in mountainous ecosystems in northwestern Iberian Peninsula.