OBJECTIVE To compare two treatment regimens associating a gonadotropin-releasing hormone agonist (GnRH-a) and human menopausal gonadotropin (hMG) for controlled ovarian hyperstimulation (COH). METHODS A prospective randomized trial. METHODS The outpatient fertility clinic of a university tertiary care center, the Hôpital A. Béclère, Clamart, France. METHODS One hundred eighty-two in vitro fertilization (IVF) candidates undergoing new or repeat IVF cycles at Hôpital A. Béclère over a 4-month period. METHODS Group 1 (7-day protocol): A short-acting preparation of GnRH-a (Tripteriline 0.1) was administered daily for 7 days, starting on cycle day 2. Ovarian stimulation with hMG was started on cycle day 4. Group 2 (long protocol): A timed release preparation of GnRH-a (Tripteriline 3.75 mg) was administered on cycle day 2. Ovarian stimulation with hMG was started after documented ovarian suppression. METHODS Response to COH, pregnancy rate (PR), tolerance. RESULTS In the 7-day protocol, the amount of hMG required was markedly lower at 24 +/- 7 than in the long protocol group requiring 42.5 +/- 9.75 vials (75 IU) (mean +/- SD). No elevation of plasma LH occurred in either group. The number of oocytes retrieved was 7.3 +/- 1 and 10.7 +/- 1.2 (mean +/- SD) in the 7-day and long protocols, respectively. Yet, the number of embryos obtained and the PRs were similar in the two treatment groups. CONCLUSIONS We observed that in COH, GnRH-a treatment could be interrupted safely several days before human chorionic gonadotropin administration without risking a premature increase of plasma luteinizing hormone. Moreover, the number of embryos available for fresh transfer and the ongoing PRs were similar in the new 7-day and in the classic long GnRH-a/hMG protocols, despite the smaller number of oocytes suggesting a greater efficiency of the 7-day protocol. The peak estradiol level and the hMG requirement were also lower in the 7-day GnRH-a/hMG protocol.