Twenty-six anovulatory women of polycystic ovarian (PCO)-type were treated with bromocriptine (Br) at a daily dose of 5 mg for 2 months. Ovulatory cycles were resumed in 18 (69.2%) women (Br-responders). No difference between pretreatment serum levels of FSH, LH, PRL and estradiol and FSH/LH ratios in Br-responders and nonresponders was observed. The geometric mean of circulating androstenedione (A-dione) in Br-responders (2.58 ng/ml) appeared higher than that in nonresponders (2.11 ng/ml) but was not statistically significant. The geometric mean of dehydroepiandtrosterone sulfate (DHEA-S) in Br-responders (1652 ng/ml) was lower (p less than 0.01) than that in nonresponders (2582 ng/ml). The ratio of DHEA-S to A-dione (D/A ratio) exhibited a highly significant between-group difference (p less than 0.001) (646 and 1222 for Br-responders and nonresponders, respectively). Br-nonresponders with high DHEA-S levels and D/A ratios tended to hyperproduction of adrenal androgen, and Br-responders with high A-dione levels and low D/A ratios to hyperproduction of ovarian androgen. The present study indicates that Br is effective in PCO-type women presumably with ovarian androgen hyperproduction. The efficacy of Br, when applied to PCO-type women, could be predicted with their D/A ratios.