We studied the peri-ovulatory and luteal phases in 38 human menopausal gonadotrophin (HMG)-stimulated cycles, in which ovulation was triggered with four different i.v. bolus ovulation triggers: 100 mu g gonadotrophin-releasing hormone (GnRH; group A, n = 9), 500 mu g GnRH agonist (GnRHa; group B, n = 10), 10 000 IU human chorionic gonadotrophin (HCG; group C, n = 10) and 500 mu g GnRH (group D, n 9), Endogenous luteinizing hormone (LH) surges occurred in all cycles of groups A, B and D. The rise was slowest but highest in group B (P < 0.0001) and lowest in group A, Although the to serum oestradiol values were similar in all groups, day +8 oestradiol and day +4 and +8 progesterone concentrations were higher in group C (P < 0.05), At day +4 and +8, serum LH concentrations were lowest (P < 0.01) but follicle stimulating hormone (FSH) concentrations were higher. Clinically, day +8 luteal scores showed a more conspicuous degree of ovarian hyperstimulation id the HCG group (P = 0.0292). Luteal insufficiency, defined as cycles with progesterone concentrations of <8 ng/ml, occurred much more frequently in groups A, B and D than in group C (day +4: P < 0.0003; day +8: P < 0.0001), despite progesterone supplementation. Three pregnancies (one in group C and two in group D) and one moderate case of ovarian hyperstimulation syndrome (OHSS) (in a nonconceptional group D cycle) occurred. These findings show that (i) ovulation occurs and pregnancy can be achieved following an endogenous LH surge induced by GnRH and its agonists, (ii) a high frequency of luteal insufficiency occurs in such cycles even with luteal supplementation and (iii) OHSS cannot be totally prevented by this approach, although cycles with an endogenous LH surge in general result in fewer subclinical signs of ovarian hyperstimulation.