Replacing GnRH agonist cotreatment for the prevention of a premature rise in LH during ovarian stimulation for in vitro fertilization (IVF) by the late follicular phase administration of GnRH antagonist may render supplementation of the luteal phase redundant, because of the known rapid recovery of pituitary function after antagonist cessation. This randomized two-center study was performed to compare nonsupplemented luteal phase characteristics after three different strategies for inducing final oocyte maturation. Forty patients underwent ovarian stimulation using recombinant (r-)FSH (150 IU/d, fixed) combined with a GnRH antagonist (antide; 1 mg/d) during the late follicular phase. When at least one follicle above 18 mm was observed, patients were randomized to induce oocyte maturation by a single injection of either r-human (h)CG (250 microg) (n = 11), r-LH (1 mg) (n = 13), or GnRH agonist (triptorelin; 0.2 mg) (n = 15). Retrieved oocytes were fertilized by either IVF or intracytoplasmatic sperm injection, depending on sperm quality. Embryo transfer was performed 3-4 d after oocyte retrieval. No luteal support was provided. Serum concentrations of FSH, LH, estradiol (E(2)), progesterone (P), and hCG were assessed at fixed intervals during the follicular and luteal phase. The median duration of the luteal phase was 13, 10, and 9 d for the r-hCG, the r-LH, and the GnRH agonist group, respectively (P = 0.005). The median area under the curve per day (from 4 d post randomization until the onset of menses) for LH was 0.50, 2.34, and 1.07 for the r-hCG, the r-LH, and the GnRH agonist group, respectively (P = 0.001). The median area under the curve per day for P was 269 vs. 41 and 16 for the r-hCG, the r-LH, and the GnRH agonist group, respectively (P < 0.001). Low pregnancy rates (overall, 7.5%; range, 0-18% per started cycle) were observed in all groups. In conclusion, the nonsupplemented luteal phase was insufficient in all three groups. In the patients receiving r-hCG, the luteal phase was less disturbed, compared with both other groups, presumably because of prolonged clearance of hCG from the circulation and the resulting extended support of the corpus luteum. Despite high P and E(2) concentrations during the early luteal phase in all three groups, luteolysis started prematurely, presumably because of excessive negative steroid feedback resulting in suppressed pituitary LH release. Hence, support of corpus luteum function remains mandatory after ovarian stimulation for IVF with GnRH antagonist cotreatment.