We performed a prospective study of 28 initially untrained college women with documented ovulation and luteal adequacy to determine whether strenuous exercise spanning two menstrual cycles would induce menstrual disorders. To ascertain the influence, if any, that weight loss might exert, we randomly assigned the subjects to weight-loss and weight-maintenance groups. Subjects were expected to run 4 miles (6.4 km) per day, progressing to 10 miles (16.1 km) per day by the fifth week, and to engage daily in 31/2 hours of moderate-intensity sports. The normalcy of the menstrual cycles during the period of exercise was judged independently according to clinical and hormonal criteria, the latter comprising serial measurements of gonadotropin and sex-steroid excretion. A higher percentage of abnormalities proved to be detectable by hormonal means (P less than 0.02). Only four subjects (three in the weight-maintenance group) had a normal menstrual cycle during training. In the weight-loss group, the number of women who had luteal abnormalities as compared with those who lost the surge in luteinizing hormone altered significantly over time, the latter occurring more frequently (P less than 0.01) as training progressed. Within six months of termination of the study, all subjects were again experiencing normal menstrual cycles. We conclude that vigorous exercise, particularly if compounded by weight loss, can reversibly disturb reproductive function in women.