Animal–vehicle collisions (AVCs) are a substantial problem in a human-dominated world, but little is known about what goes wrong, from the animal’s perspective, when a collision occurs with an automobile, boat, or aircraft. Our goal is to provide insight into reactions of animals to oncoming vehicles when collisions might be imminent. Avoiding a collision requires successful vehicle detection, threat assessment, and evasive behaviour; failures can occur at any of these stages.Vehicle detection seems fairly straightforward in many cases, but depends critically on the sensory capabilities of agiven species. Sensory mechanisms for detection of collisions (looming detectors) may be overwhelmed by vehicle speed. Distractions are a likely problem in vehicle detection, but have not been clearly demonstrated in any system beyond human pedestrians. Many animals likely perceive moving vehicles as non-threatening, and may generally be habituated to their presence. Slow or minimal threat assessment is thus a likely failure point in many AVCs, but this is not uniformly evident. Animals generally initiate evasive behaviour when a collision appears imminent, usually employing some aspect of native antipredator behaviour. Across taxa, animals exhibit a variety of behaviours when confrontedwith oncoming vehicles. Among marine mammals, right whales Eubalaena spp., manatees Trichechus spp., and dugongs Dugong dugon are fairly unresponsive to approaching vehicles, suggesting a problem in threat assessment. Others, such as dolphins Delphinidae, assess vehicle approach at distance. Little work has been conducted on the behavioural aspectsof AVCs involving large mammals and automobiles, despite their prevalence. Available observations suggest that birds do not usually treat flying aircraft as a major threat, often allowing close approach before taking evasive action, as they might in response to natural predators. Inappropriate antipredator behaviour (often involving immobility) is a major source of AVCs in amphibians and terrestrial reptiles. Much behavioural work on AVCs remains to be done across awide variety of taxa. Such work should provide broad phylogenetic generalizations regarding AVCs and insights intomanaging AVCs.