OBJECTIVE: To determine whether metformin would safely reduce the rate of first-trimester spontaneous abortion without teratogenicity in 19 women with the polycystic ovary syndrome (PCOS). DESIGN: Prospective pilot study. SETTING: Outpatient. PATIENT(S): Twenty-two previously oligoamenorrheic, nondiabetic women with PCOS; 125 women with PCOS who were not currently pregnant and who had > or = 1 previous pregnancy while they were not receiving metformin. INTERVENTION(S): Metformin, 1.5-2.55 g/day, throughout pregnancy. MAIN OUTCOME MEASURE(S): Rates of first-trimester spontaneous abortion and teratogenicity. RESULT(S): Before metformin, 10 women had 22 previous pregnancies with 16 first-trimester spontaneous abortions (73%). While receiving metformin, these 10 women had 6 normal live births (60%), 1 spontaneous abortion (10%), and 3 normal ongoing pregnancies (30%) (all > or = 13 weeks; median gestation, 23 weeks). Among women receiving metformin, including those with live births and normal pregnancy for at least the first trimester, 1 of 10 (10%) had first-trimester spontaneous abortion compared with 73% in 22 previous pregnancies without metformin (P