This study was carried out to investigate the relationship between the level of training and the impairment of ovarian function among female recreational distance runners, and its reversibility. Thirty-six female distance runners self-recorded, for 7 consecutive days each month, the duration and distance of daily running from October 1989 to May 1990 in Great Britain. During the last 3 months of the survey, saliva samples were taken for progesterone assay and a subsample was measured for body composition. No trend in weight loss was observed over the season of training. Amenorrheic (AM) and oligomenorrheic (Oligo) runners had a significantly lower body mass index than eumenorrheic (EU) and irregularly menstruating (IM) runners. The amenorrheic and oligomenorrheic subjects did not show any rise in progesterone, at any time, during the 3 months of sampling. The eumenorrheic subjects showed evidence of a rise in progesterone, though the mean level was always significantly lower than that of sedentary controls. The most severely impaired runners (AM and Oligo) ran more than EU or irregularly menstruating runners in this sample, had lower body weight, a younger age and had a significantly lower body mass index (BMI). They tended to run faster during training sessions than those with apparently normal menstrual cycle or just irregular periods. It is suggested that low BMI, which is an indicator of body energy stores, reflects the intensity of regular training runs among female athletes with a stable body weight. It is possible that repeated elevations of beta-endorphins or other suppressors of gonadotropin release, secreted above a level of training commonly exceeded by long distance runners, when concurrent with energy restriction, could contribute to impairment of menstrual cycle.