<p>Micro-endemic animals face high extinction risks. Species distribution models offer powerful tools for effective conservation strategies, but their implementation is hindered by the resolution of environmental data such as land cover. Here, we assessed the efficacy of one regional versus two continental land cover datasets in predicting habitat suitability for <i>Salamandra atra aurorae</i>, a fully terrestrial amphibian endemic to a ca. 30 km² area in Northern Italy. We built three species distribution models with the same spatial resolution of 100 × 100&#xa0;m using the same topographic and climatic predictors but varying the land cover dataset describing forest classes. We used a composite regional dataset assembled from local sources, the Corine Land Cover and the Sentinel-2 Global Land Cover, and compared their capacity to identify the ecological requirements of the species. The models performed comparably, identifying elevation, temperature, and tree composition as primary drivers of habitat suitability and predicting similar suitable areas. However, while all models recognized coniferous forests as more suitable than broadleaf forests, only the land cover classification of the regional dataset allowed to identify different suitability among coniferous forests. Notably, the model using the regional dataset identified old-growth stands with <i>Abies alba</i> as the most suitable, aligning with previous ecological studies. Our case study highlights the limitations of widely used continental land cover datasets in recognising key environmental features influencing habitat suitability for a micro-endemic animal. We showed that incorporating regional land cover data can enhance the accuracy of species distribution models providing more detailed ecological information to guide conservation efforts.</p>