In brief: Most information about running injuries comes from case reports. Although useful, the case series does not take into consideration the population from which the injuries arose and is therefore an inappropriate method on which to base causal inference. The epidemiological method is a more powerful approach because, by definition, it takes into account the population from which the injuries arose. A review of three epidemiological studies shows that the only reasonably well-established cause of running injuries is the number of miles run per week. More information is needed to establish the relationship between injury and characteristics of the runner, characteristics of running, and characteristics of the running environment. More research on the causes of running injuries is needed and should be directed to those factors over which the runner has control.