Ten hirsute women with polycystic ovarian syndrome (PCO) and nine with idiopathic hirsutism (IH) underwent selective ovarian suppression with leuprolide for 5-6 months and then were randomized to receive, in addition, dexamethasone or placebo for 4 more months. Serum hormone levels and hair growth rates were determined before and after each treatment period. During the initial treatment period with leuprolide alone, testosterone decreased by 54 +/- 6% (mean +/- SEM) in PCO and by 36 +/- 3% in IH (P = 0.02). Androstenedione decreased by 53 +/- 6% in PCO and by 31 +/- 7% in IH (P = 0.02). Androstanediol glucuronide (Adiol-G) decreased by 14 +/- 6% in PCO and by 7 +/- 3% in IH. There was no change in dehydroepiandrosterone sulfate (DHEAS). While initial serum androgen levels were higher in PCO than in IH, they were similar after ovarian suppression in the two groups. After ovarian suppression, Adiol-G was more consistently correlated with testosterone and androstenedione than was DHEAS, suggesting that Adiol-G may be a better marker than DHEAS of adrenal androgen secretion. Hair growth rates decreased by 37 +/- 6% in PCO and by 14 +/- 10% in IH (P = 0.07). The change in hair growth correlated with the change in androstenedione (r = 0.66; P = 0.002), but not significantly with the change in testosterone (r = 0.29; P = 0.2). After the addition of dexamethasone therapy (0.5 mg daily), testosterone, androstenedione, and DHEAS levels fell to near or below assay detection limits, while Adiol-G decreased by 80 +/- 3%. Hair growth rates decreased slightly more in women during dexamethasone (46 +/- 6%) than during placebo (26 +/- 9%; P = 0.18). In summary, the ovary was the major source of circulating testosterone and androstenedione in PCO. The adrenal contributed a substantial minority of these hormones in PCO and was the major source of androgen secretion in IH. Adrenal hyperandrogenism was common in both IH and PCO. Hair growth rates correlated best with changes in serum androstenedione levels. Adiol-G, which was derived primarily from adrenal precursors, was a better marker of adrenal androgen secretion than was DHEAS in these subjects.