Menstrual irregularity is a common complaint at presentation in women with Cushing's syndrome, although the etiology has been little studied. We have assessed 45 female patients (median age, 32 yr; range, 16 -41 yr) with newly diagnosed pituitary-dependent Cushing's syndrome. Patients were subdivided into 4 groups according to the duration of their menstrual cycle: normal cycles (NC; 26 -30 days), oligomenorrhea (OL; 31-120 days), amenorrhea (AM; Ͼ120 days), and polymenorrhea (PM; Ͻ26 days). Blood was taken at 0900 h for measurement of LH, FSH, PRL, testosterone, androstenedione, dehydroepiandrosterone sulfate, estradiol (E 2 ), sex hormone-binding globulin (SHBG), and ACTH; cortisol was sampled at 0900, 1800, and 2400 h. The LH and FSH responses to 100 g GnRH were analyzed in 23 patients. Statistical analysis was performed using the nonparametric Mann-Whitney U and Spearman tests.Only 9 patients had NC (20%), 14 had OL (31.1%), 15 had AM (33.3%), and 4 had PM (8.8%), whereas 3 had variable cycles (6.7%). By group, AM patients had lower serum E 2 levels (median, 110 pmol/L) than OL patients (225 pmol/L; P Ͻ 0.05) or NC patients (279 pmol/L; P Ͻ 0.05), and higher serum cortisol levels at 0900 h (800 vs.602 and 580 nmol/L, respectively; P Ͻ 0.05) and 1800 h (816 vs. 557 and 523 nmol/L, respectively; P Ͻ 0.05) and higher mean values from 6 samples obtained through the day (753 vs. 491 and 459 nmol/L, respectively; P Ͻ 0.05). For the whole group of patients there was a negative correlation between serum E 2 and cortisol at 0900 h (r ϭ Ϫ0.50; P Ͻ 0.01) and 1800 h (r ϭ Ϫ0.56; P Ͻ 0.01) and with mean cortisol (r ϭ Ϫ0.46; P Ͻ 0.05). No significant correlation was found between any serum androgen and E 2 or cortisol. The LH response to GnRH was normal in 43.5% of the patients, exaggerated in 52.1%, and decreased in 4.4%, but there were no significant differences among the menstrual groups. No differences were found in any other parameter.In summary, in our study 80% of patients with Cushing's syndrome had menstrual irregularity, and this was most closely related to serum cortisol rather than to circulating androgens. Patients with AM had higher levels of cortisol and lower levels of E 2 , while the GnRH response was either normal or exaggerated. Our data suggest that the menstrual irregularity in Cushing's disease appears to be the result of hypercortisolemic inhibition of gonadotropin release acting at a hypothalamic level, rather than raised circulating androgen levels. (J Clin Endocrinol Metab 83: 3083-3088, 1998)