In 43 amenorrheic women with polycystic ovary syndrome (PCOS), 31 (74%) with fasting hyperinsulinemia (> or =20 microU/mL), our aim was to determine whether Metformin (Bristol-Myers Squibb, Princeton, NJ), which reduces hyperinsulinemia, would reverse the endocrinopathy of PCOS, allowing resumption of regular normal menses. A second aim was to assess the effects of weight loss versus other Metformin-induced effects on ovarian function, and to determine if there were different responses to Metformin between those who lost weight and those who did not. A third aim was to assess associations between PCOS, 4G/5G polymorphism in the promoter sequence of the plasminogen activator inhibitor-1 gene (PAI-1 gene), and PAI activity (PAI-Fx). Of the 43 women, 40 (93%) had normal fasting blood glucose and 37 had normal hemoglobin A1C (HgA1C); onlythree (7%) had type 2 diabetes mellitus. Metformin (1.5 to 2.25 g/d) was given for 6.1+/-5.1 months (range, 1.5 to 24), to 16 patients for less than 3 months, to 12 for 3 to 6 months, and to 15 for at least 6 months. On Metformin, 39 of 43 patients (91%) resumed normal menses. The percentage of women resuming normal menses did not differ among treatment duration groups (P<.1) or among dose groups (P>.1). The body mass index (BMI) decreased from 36.4 + 7 Kg/m2 at study entry to 35.1+/-6.7 on Metformin (P=.0008). Of 43 patients, 28 (67%) lost weight (1 to 69 pounds), with nine (21%) losing at least 12 pounds. On Metformin, the median fasting serum insulin decreased from 26 microU/mL to 22 (P=.019), testosterone decreased from 61 ng/dL to 47 (P=.003), and estradiol increased from 41 pg/mL to 71 (P=.0001). Metformin-induced improvements in ovarian function were independent of weight loss (testosterone decrease, P<.002; estradiol increase, P<.0004). The change in response variables on Metformin did not differ (P>.05) between those who lost weight and those who did not, excepting Lp(a), which increased 4 mg/dL in those who lost weight and decreased 9 mg/dL in those who did not (P = .003). The change in response variables on Metformin did not differ among the five quintiles of weight loss, excepting fasting glucose (P<.05), which increased 6 mg/dL in those who lost the least weight on Metformin versus those in the 60th to 80th percentile for weight loss, in whom glucose decreased 33 mg/dL. Although the pretreatment fasting serum insulin was not significantly correlated with testosterone (r=.24, P=.13) or androstenedione (r=.27, P=.09), on Metformin, the change in insulin correlated positively with the change in testosterone (r=.35, P=.047) and with the change in androstenedione (r=.48, P=.01). Patients were more likely than normal controls (83% v 64%, P=.016) to be heterozygous or homozygous for 4G polymorphism of the PAI-1 gene and were also more likely to have high PAI-Fx (> or =22 U/mL, 28% v3%, chi(2)=10.1, P=.001). Metformin reduces the endocrinopathy of PCOS, allowing resumption of normal menses in most (91%) previously amenorrheic women with PCOS.