OBJECTIVE: To determine whether basal cortisol levels are elevated in exercise-associated amenorrhea. DESIGN: Survey, with hormone levels measured weekly for 1 month and patients followed clinically for 6 months. SETTING: Volunteers were recruited through media advertisements and fliers. PARTICIPANTS: Ninety-two women were enrolled; 71 (77%) completed the study. Subjects were grouped by menstrual and activity histories reported by a self-administered questionnaire. After 6 months, final groups were assigned: amenorrheic athletes, 19; eumenorrheic athletes, 35; a transition group of amenorrheic athletes who had resumed menses after entering the study, 7; and normal cyclic nonathletes, 10. INTERVENTIONS: Four weekly resting blood samples (0800 to 1000 hours) were obtained and measured for cortisol, estradiol, progesterone, and prolactin levels. Lumbar bone mineral density was measured by dual-photon densitometry. MEASUREMENTS and MAIN RESULTS: Mean (+/- SE) cortisol levels were higher in amenorrheic athletes (585 +/- 33 nmol/L) than in eumenorrheic athletes (411 +/- 14 nmol/L), transition athletes (378 +/- 33 nmol/L), or nonathletic women (397 +/- 30 nmol/L) (P less than 0.01). Of nine women with abnormally high cortisol levels (greater than 579 nmol/L), eight were amenorrheic athletes, and one was a eumenorrheic athlete. Bone mineral density was lower in amenorrheic athletes than in the other three groups (P less than 0.01). CONCLUSIONS: Increased glucocorticoid levels may be an etiologic factor in exercise-associated amenorrhea. High cortisol levels could also contribute to decreased bone density. The failure of amenorrheic athletes with hypercortisolemia to regain menses within 6 months suggests that they are at risk for a prolonged acyclic state.