The aim of this 12-month prospective study was to investigate risk factors for stress fractures in a cohort of 53 female and 58 male track and field athletes, aged 17 to 26 years. Total bone mineral content, regional bone density, and soft tissue composition were measured using dual-energy x- ray absorptiometry and anthropometric techniques. Menstrual characteristics, current dietary intake, and training were assessed using questionnaires. A clinical biomechanical assessment was performed by a physical therapist. The incidence of stress fractures during the study was 21.1%, with most injuries located in the tibia. Of the risk factors evaluated, none was able to predict the occurrence of stress fractures in men. However, in female athletes, significant risk factors included lower bone density, a history of menstrual disturbance, less lean mass in the lower limb, a discrepancy in leg length, and a lower fat diet. Multiple logistic regression revealed that age of menarche and calf girth were the best independent predictors of stress fractures in women. This bivariate model correctly assigned 80% of the female athletes into their respective stress fracture or nonstress fracture groups. These results suggest that it may be possible to identify female athletes most at risk for this overuse bone injury.