Aims: To identify visual and medical risk factors for motor vehicle collisions (MVCs). Methods: Data from four cohorts of older drivers from three states were pooled (n = 3158). Health information was collected at baseline, and MVC data were obtained prospectively. Cox proportional hazards regression was used to estimate rate ratios (RRs) and 95% CIs for associations between medical characteristics and MVCs. Results: A total of 363 MVCs were observed during the study period (1990-1997), of which 145 were at fault, and 62 were injurious. Falls and impaired useful field of view (UFOV) were positively associated with overall MVCs. At-fault MVCs were also positively associated with falls and UFOV impairment, and inversely with cancer. Injurious MVCs were positively associated with arthritis and neurological disease, and inversely with hypertension. Conclusions: These findings show similarities and differences across the risk factors for all, at-fault and injurious MVCs, and point to the need for verification and possible interventions.