Objective To investigate the effect of metformin used during early pregnancy on pregnancy outcome of pregnant women with polycystic ovary syndrome (PCOS).Methods Two hundred and fifteen pregnant women with PCOS history were selected into our investigation from March 2007 to February 2010,among which,58 patients were taken as metformin group as they had taken metformin during early pregnancy,or with impaired glucose tolerance complicated by marked hyperinsulinemia and history of recurrent miscarriage.All patients in the study group received metformin (1.0～1.5 g/d) during early pregnancy.The rest 157 gravidas with PCOS history did not receive metformin as control group.The outcomes of pregnancy in the two groups were compared with x2 or t test.Results The early spontaneous miscarriage rate of metformin group was significantly lower than that of control group (5.2 ％ vs 28.7 ％,x2 =13.476,P＜0.01).There was no statistical difference in the rate of late abortion (6.90％ vs 6.4 ％ ),hypertension during pregnancy (3.9％ vs 4.9％),preeclampsia (3.9％ vs 3.9％),preterm birth (13.7％ vs 20.6％) and oligohydramnios (5.9％ vs 3.0％) between control group and metformin group (P ＞ 0.05 respectively).And there was no significant difference in neonatal birth weights [(2925±530) g vs (2910± 659) g],the rate of asphyxia (3.7％ vs 6.8％),malformation (0.0％ vs 0.9％),hypoglycaemia ( 5.6％ vs 6.8％),fetal macrosomia ( 1.9％ vs 3.4％),fetal growth restriction (5.6％ vs 6.0％) and rate of neonatal intensive care unit admission (18.0％ vs 27.4％) between the two groups (P＞0.05,respectively).The infants were followed up for 3 years and all are physically and mentally normal.Conclusions Metformin administration during early pregnancy might reduce the early spontaneous miscarriage rate of PCOS patients,and the adverse effects of metformin have not been identified on fetals and infants.