We have found that women with typical polycystic ovary syndrome have supranormal plasma 17-hydroxyprogesterone responses to a 100-micrograms test dose of the gonadotropin-releasing hormone agonist nafarelin without evidence of hindered estrogen secretion. To understand the basis of this response, we computed the apparent efficiency of the steps in steroid biosynthesis from the pattern of plasma steroids in response to nafarelin. The proximate cause appears to be excessive 17 alpha-hydroxylase activity and high, yet partially down-regulated, 17,20-lyase activity in the delta 4-pathway. These results suggest that this pattern of steroid secretion results from abnormal regulation (dysregulation) of these activities, possibly involving the enzyme cytochrome P450c17. To determine the usefulness of nafarelin testing for the diagnosis of ovarian hyperandrogenism, we then prospectively studied 40 hyperandrogenic women. The plasma 17-PROG response to nafarelin was supranormal in 58% of the women. The responses of 17-PROG to nafarelin and free testosterone to dexamethasone correlated well and were concordant in approximately 85% of cases. Baseline serum luteinizing hormone concentration was elevated in only 48% of cases. To understand ovarian structure-function relationships, we studied another 20 consecutive hyperandrogenic women. Among seven women with polycystic ovaries, five had an elevated LH level, and four of these five (80%) had an elevated 17-PROG response to nafarelin. Conversely, about half of patients with the PCOS-like disorder of ovarian function did not have polycystic ovaries. Ovarian stromal area, but not LH levels, correlated significantly (r = 0.45) with the 17-PROG response to nafarelin. Thus, both stromal hyperplasia and dysregulation of steroidogenesis seem to be manifestations of abnormal intraovarian regulation of cell growth and function. We conclude that a PCOS-like disorder of ovarian function in response to nafarelin testing is found in approximately half of hyperandrogenic women. The pathogenetic implication of our results is that abnormal intraovarian modulation of LH action seems to be a major factor in ovarian hyperandrogenism. The diagnostic implication of our data is that ovarian androgen excess will often be missed by use of common diagnostic criteria for PCOS.