Cost surface (CS) models have emerged as a useful tool to examine the interactions between landscapes patterns and wildlife at large-scale extents. This approach is particularly relevant to guide conservation planning for species that show vulnerability to road networks in human-dominated landscapes. In this study, we measured the functional connectivity of the landscape in southern Portugal and examined how it may be related to stone marten road mortality risk. We addressed three questions: (1) How different levels of landscape connectivity influence stone marten occurrence in montado patches? (2) Is there any relation between montado patches connectivity and stone marten road mortality risk? (3) If so, which road-related features might be responsible for the species' high road mortality? We developed a series of connectivity models using CS scenarios with different resistance values given to each vegetation cover type to reflect different resistance to species movement. Our models showed that the likelihood of occurrence of stone marten decreased with distance to source areas, meaning continuous montado. Open areas and riparian areas within open area matrices entailed increased costs. We found higher stone marten mortality on roads in well-connected areas. Road sinuosity was an important factor influencing the mortality in those areas. This result challenges the way that connectivity and its relation to mortality has been generally regarded. Clearly, landscape connectivity and road-related mortality are not independent. ?? 2010 Springer-Verlag.