Several studies have indicated that ovulation induction with human menopausal gonadotropin (hMG)/human chorionic gonadotropin (hCG) or clomiphene citrate (CC) is associated with luteal phase defect. To assess the efficiency of luteal support by hCG to an infertile population undergoing ovulation induction, with CC/hCG or hMG/hCG, we have randomly administered 2500 IU hCG intramuscularly on days 3, 6, and 9 after ovulation induction by 10,000 IU of hCG to 74 patients on 265 treatment cycles. As controls served 357 ovulation induction cycles in the same 74 patients. The treatment cycles were randomly alternated with control cycles so that each patient served as her own control. However, the mean +/- standard deviation (SD) midluteal P was 38.1 +/- 10.8 ng/ml in the study group versus 15.7 +/- 10.5 ng/ml in the control group (P less than 0.001). Luteal phase length was 15.4 +/- 1.5 days in the treatment group versus 12.1 +/- 1.7 in the control group (P less than 0.01). In the treatment group, 64.8% of the patients achieved pregnancy (27% pregnancies/treatment cycle) versus 47.3% in the control group (11.5% pregnancies/control cycle) (P less than 0.01). The pregnancy wastage rates (including abortions and "chemical" pregnancies) were 30.6% in the treatment group versus 56% in the control group (P less than 0.01). We conclude that repetitive hCG administration may be an efficient luteal support in infertile patients undergoing ovulation induction.