OBJECTIVE: To examine the possible impact of abnormal adrenal steroidogenesis on the ovarian dysfunction seen in polycystic ovarian disease (PCOD). DESIGN: Prospective analysis of blood sampling monthly for 6 months, then three times weekly for 90 days. SETTING: Tertiary institutional outpatient care. PARTICIPANTS: Six anovulatory women with a diagnosis of PCOD. INTERVENTION: Six-month suppression with gonadotropin-releasing hormone agonist (GnRH-a) followed by suppression with dexamethasone (DEX) for 90 days. MAIN OUTCOME MEASURES: Serum levels of testosterone (T), androstenedione (A), dehydroepiandrosterone (DHEA), dehydroepiandrosterone sulfate (DHEAS), cortisol, estradiol (E2), progesterone (P), follicle-stimulating hormone (FSH), luteinizing hormone (LH) and bioactive LH. RESULTS: Gonadotropin-releasing hormone agonist administration suppressed greater than 60% of the circulating levels of T and A, suggesting an ovarian origin. Minimal changes of DHEA, DHEAS, and cortisol were seen. With the addition of DEX, there was greater than 90% suppression of the total circulating A, T, DHEA, DHEAS, and cortisol, supporting the adrenal origin of the non-GnRH-a suppressible androgens. Excessive ovarian T and A secretion returned during the 90-day recovery study period in spite of rises of FSH concentrations that changed the ratio of FSH to LH in all subjects. Four of the six women failed to ovulate. In comparison of the women who did and did not ovulate during recovery, no differences in absolute levels or changes in concentrations of steroids or gonadotropins could be detected. CONCLUSIONS: Using sequential and simultaneous administration of GnRH-a and DEX, we were able to delineate the contributions of the ovaries and adrenals to the abnormal steroidogenesis seen in PCOD. Despite prolonged suppression of ovarian and then adrenal steroidogenesis, ovarian dysfunction, evidenced by abnormal androgen production, returned with cessation of agonist administration.