The association of menstrual dysfunction of athletes with changes in body composition has been controversial, because most estimations of body fatness have been indirect. Using magnetic resonance imaging, we quantified the sc and internal fat over a specific volume from the fifth thoracic vertebra to femoral fat in the upper thigh and at 4 other anatomical landmarks of 17 athletes (13 oarswomen and 4 runners) compared to that in 11 nonathletic controls. The magnetic resonance imaging data were also analyzed for the athletes and controls in relation to ovulatory status, which was determined by assay of urinary pregnanediol glucuronide, and in relation to the extent of 2-hydroxylation of estradiol to a nonpotent metabolite, 2-hydroxyestrone, which was evaluated by radiometric analysis. We found that 1) the relative and absolute body fat values of the athletes were significantly less (P < 0.05) than those of the controls overall and at each of the six regional sites, although the body weights of the rowers were significantly heavier than those of the controls, and the runners did not differ from the controls; 2) the ratio of sc fat to internal fat was 80%:20% among both athletes and controls, even though the athletes had significantly less fat; 3) the extent of estradiol 2-hydroxylation was significantly (P = 0.005) inversely related to total fat as a percentage of the total volume and to sc fat as a percentage of the total volume (P = 0.004) overall and at each of the regional fat depots; 4) athletes with menstrual disorders had significantly decreased sc and internal fat overall and at all regional sites compared to controls; and 5) a subgroup of ovulatory rowers had an apparent increase or lack of decrease in internal fat at the level of vertebrae lumbar 4, sacral 1, and sacral 4, compared to controls, whereas their sc fat was decreased at these sites compared to that in controls. Changes in regional fat deposits of both sc and internal fat may be involved in the menstrual dysfunction of the athletes in addition to their decreased overall fatness. The body weight and body mass index of well trained athletes can be a misleading index of body composition.