The purpose of this study was to determine whether the use of clomiphene results in a higher incidence of spontaneous abortion than occurs naturally in subfertile patients. Reproductive outcomes of 1744 clomiphene pregnancies were compared to outcomes of 3245 spontaneous pregnancies in a prospective study. Abortion was classified as clinical if a sac was seen on ultrasound or if it occurred after 6 gestational weeks, and as preclinical if a quantitative human chorionic gonadotrophin (HCG) was > or = 25 mIU and no sac was seen or abortion occurred earlier. The overall incidence of abortion was higher for clomiphene pregnancies (23.7%), compared with spontaneous pregnancies (20.4%) (P < 0.01). Preclinical abortions were increased by clomiphene for all ages (5.8 versus 3.9%, P < 0.01) and for age > or = 30 years (8.0 versus 4.9%, P < 0.001), but not for age < 30 years (3.7 versus 3.0%). Clinical abortions were increased by clomiphene for age < 30 years (15.9 versus 11.2%) (P < 0.01), but not for age > or = 30 years (20.1 versus 22.3%) or all ages (18.0 versus 16.4%). Clinical abortions occurred 22% less often following clomiphene compared with spontaneous pregnancies for patients with luteal insufficiency (18.3 versus 23.6%, P < 0.05). We conclude that the increase in abortion due to clomiphene is small and may be related to different causes for women aged < 30 and > or = 30 years, and also that clomiphene may decrease clinical abortions in patients with luteal insufficiency.