Road mortality is among the most visible and well-documented impacts of transportation infrastructure on wildlife, with millions of animals killed annually across a wide range of taxonomic groups. Each roadkill record results from the combination of two independent processes: the actual collision between an animal and a vehicle and the subsequent detection of the carcass by an observer. In this chapter, we synthesize current knowledge into two complementary conceptual frameworks: one focused on roadkill risk and the other on detection biases in road-mortality surveys. In the first framework, roadkill probability emerges from the interaction between species exposure to roads and its susceptibility to collisions. In the second framework, we outline how three sequential processes (carcass-location bias, carcass-persistence bias, and carcass-observation bias) can cause significant discrepancies between observed and actual mortality. We conclude by identifying key research priorities, including the urgent need to expand taxonomic and geographic coverage, incorporate species traits, and apply emerging technological and statistical tools. Embedding roadkill studies within broader ecological and policy contexts is crucial to improve mitigation measures, guide conservation planning, and support biodiversity and sustainability goals. Collectively, these advances provide a roadmap for a more integrative, accurate, and policy-relevant research agenda on wildlife road mortality.