Pulsatile GnRH administration for induction of ovulation is often ineffective in polycystic ovarian disease (PCOD) patients. To clarify and correct the endocrine mechanisms underlying this deranged response we gave pulsatile GnRH (5 micrograms, iv, every 60 min) to idiopathic hypogonadotropic hypogonadism (IHH) patients with primary amenorrhea for 19 cycles and to PCOD patients for 24 cycles before (pre-A) and for 25 cycles after (post-A) GnRH analog suppression. Compared to IHH, pre-A cycles were characterized by elevated LH, estradiol, and testosterone; reduced luteal phase progesterone; and low ovulatory (38%) and pregnancy rates (8%). Conversely, LH, estradiol, and follicular phase testosterone levels were lower in post-A than in pre-A cycles, while luteal phase progesterone was higher; the endocrine pattern of post-A cycles closely resembled the one of IHH cycles. The ovulatory and pregnancy rates of PCOD patients improved remarkably in post-A cycles (90% and 38%, respectively). Excessive body weight was associated with a lower incidence of ovulation in both pre-A (15%) and post-A cycles (75%). A worse endocrine pattern and a lower ovulatory rate (50%) were obtained when a second consecutive post-A cycle occurred without repeating GnRH analog suppression. No signs of even mild ovarian hyperstimulation and no multiple pregnancies were recorded in the post-A cycles. We conclude that in PCOD 1) deranged pituitary sensitivity, excessive ovarian androgen secretion, and obesity critically affect folliculogenesis and ovulation; 2) pituitary-gonadal suppression with a GnRH analog markedly improves the endocrine and clinical responses to pulsatile GnRH ovulation induction; 3) optimal results can be achieved only when each pulsatile GnRH cycle is preceded by GnRH analog suppression; and 4) pulsatile GnRH is highly effective and safe for ovulation induction, provided that PCOD subjects are pretreated with a GnRH analog.