Collisions between vehicles and wildlife pose significant threats to terrestrial mammal populations, particularly in tropical dry forests characterized by high endemism and fragmentation. Understanding spatial patterns of road mortality and connectivity constraints remains limited in these ecosystems. Here, we conducted a two-year systematic survey of mammal roadkill along an 83-km highway segment in the Tumbesian dry forest of southwestern Ecuador, recording 91 individuals across nine species. We integrated roadkill data with circuit-theory connectivity modeling and spatial prioritization analyses to identify key areas for conservation and restoration. Landscape structure metrics did not predict roadkill rates, but species-specific traits influenced vulnerability. Our spatial framework reveals discrete priority zones where road mortality may disrupt functional connectivity. These findings demonstrate that empirical mortality data, when integrated with connectivity modeling, provide a non-redundant diagnostic tool for targeted mitigation, essential for maintaining biodiversity in increasingly fragmented tropical dry forests.