Past work identifying "accident-prone" children with disproportionate rates of injury has been based on clinical data gathered in office, clinic, or emergency department settings. To avoid the biases inherent in such designs, we studied recurrent injuries in a school district population, utilizing a prospective surveillance system to identify injuries meeting standardized criteria. During three school years we observed 54,874 students, ranging in age from 6 to 18 years, for recurrent injuries, which were defined as those occurring in a year in which two or more injuries were reported for the child. Five hundred seventy-three recurrently injured children (1% of the school district population) sustained 1405 injuries (17% of the overall injury experience), a proportion significantly but only slightly greater than that expected on the basis of chance alone (14%). Most recurrently injured children were injured in only a single year, with only 15 children sustaining injuries in all three study years. Age, sex, and type of school were significant correlates of recurrent injury rates, with junior high school students, boys, and students attending schools with alternative educational programs having the highest rates of injury recurrence. The findings indicate that (1) a small group of schoolchildren sustain a disproportionate share of the overall injury experience, (2) the majority of recurrently injured children experience only transient periods of enhanced injury risk, and (3) preventive strategies may benefit from investigation of developmental and social environmental factors that alter such risk.