The pathogenesis of hypothalamic progestin-nonresponsive amenorrhea is unclear and this disease often fails to respond to treatment. The pulsatile patterns of diurnal and nocturnal secretion of serum LH as well as serum levels of melatonin were examined to improve the understanding of the pathogenesis and to develop strategies for the management of a severe type of hypothalamic amenorrhea. Four types of LH pulsatile patterns were observed: a) no pulse during the day or night (Group 1); b) more than 1 pulse only at night (Group 2); c) only 1 pulse during the day and more than 2 pulses at night (Group 3); and d) more than 2 pulses during the day and at night (Group 4). Serum estradiol was less than 30 pg/mL, and the serum PRL and PRL response to TRH did not differ among the four groups. The basal level and the pulse amplitude of LH increased successively from Group 1 to Group 4. The serum level of melatonin at night was noticeably increased in Group 1 and correlated negatively with the LH pulse frequency at night. After 6-month hormone replacement therapy with estrogen and progesterone, the rate of improvement in ovarian function were 0%, 33.3%, 57.1% and 67.0% in Groups 1, 2, 3, and 4, respectively. In 5 patients, the LH pulse pattern was re-examined at 6 months, the LH pulsatile pattern was changed from that of Group 1 to that of Group 4, with a decrease in serum concentrations of nocturnal melatonin, indicating improved ovarian function. In conclusion, classification of patients according to the LH secretion pattern is useful in establishing the severity of hypothalamic disturbance in hypothalamic progestin-nonresponsive amenorrhea and in predicting its prognosis; in addition nocturnal melatonin can be used as a marker for severer cases of hypothalamic amenorrhea.