OBJECTIVE:The aim was to monitor carefully follicular growth arrest in polycystic ovaries by assay of hormones in individual follicles. DESIGN AND PATIENTS:Fluid from follicles less than or equal to 10 mm was obtained from ovaries of 16 regularly cycling women between days 1 and 12 of the follicular phase (controls, n = 120 follicles), polycystic ovaries of five women with polycystic ovary syndrome (n = 43), and polycystic ovaries from 14 long-term testosterone treated female to male transsexuals (n = 120). MEASUREMENTS:Fluid was assayed for oestradiol, androstenedione, and immunoactive inhibin. Luteinizing hormone, follicle-stimulating hormone, and testosterone levels were estimated in serum. RESULTS:Median serum LH was lower in transsexuals than in controls (P less than 0.05), and in polycystic ovary syndrome (P less than 0.01). Median serum testosterone was not significantly different between polycystic ovary syndrome and transsexuals, and was elevated in both groups as compared to controls (P less than 0.01). Oestradiol was present in all follicles obtained from polycystic ovaries of polycystic and transsexual patients, in which no follicle greater than 10 mm could be detected. In the three groups, between-patient differences in mean oestradiol, androstenedione, inhibin, and androstenedione/oestradiol ratio were significantly larger than expected in view of the variation between follicles within individuals. Taking into account this between-patient difference, no significant differences could be established between the three groups for all endocrine parameters. The percentage of presumed healthy follicles (androstenedione/oestradiol ratio less than or equal to 4) was 12% in controls, 17% in polycystics, and 14% in transsexuals, and was not significantly different between groups. CONCLUSIONS:The results may indicate that (1) abnormally high circulating androgen concentrations with or without elevated LH levels disturb the process of selection, and could therefore play a role in the pathogenesis of polycystic ovaries; (2) in polycystic ovaries from polycystic ovary syndrome and transsexual patients, aromatase activity is present in vivo in small antral follicles, and the proportion of presumed healthy follicles is not different from that encountered in normal ovaries; (3) oestradiol levels are not different between non-dominant follicles of normal and polycystic ovaries, suggesting that only enhancement of aromatase activity by FSH may be disrupted in polycystic ovaries, (4) because androstenedione levels are not different comparing follicles of normal and polycystic ovaries, hyperandrogenaemia in the syndrome seems to originate from the abnormally high number of cystic atretic follicles generally observed in polycystic ovaries; (5) marked variation in the endocrine follicular microenvironment within and between-women precludes pooling fluid from several follicles.