Hazardous roadway locations can be identified by monitoring the distribution of collisions on roads or by community concern following spectacular collisions. An in-depth investigation of collisions at a suspected high-risk site can allow engineers to conclude about causative factors. Expensive roadway investigations and modifications, however, should be undertaken only after an excess risk is demonstrated. We suggest using standard analytic cohort study methods for assessing suspected high-risk locations. The proposed approach complements the traditional approach of engineers, but differs from other epidemiologic studies of roadway hazards. This new application of standard techniques could help optimize the allocation of resources. (Epidemiology 1990;1:311–314)