To elucidate the clinical background of the luteal-phase defect (LPD), 201 LPD cycles were studied in 753 infertile women. One hundred and twenty-one cases (62.2%) of LPD showed transient hyperprolactinemia. In transient hyperprolactinemia, there was a significant inverse correlation between serum prolactin (PRL) 30 min after the 500-µg intravenous loading of thyrotropin-releasing hormone TRH (PRL<sub>30</sub>) and progesterone (P<sub>4</sub>) in the luteal phase (r=-0.67, p < 0.005). Mature follicles (diameter > 20 mm as determined by ultrasonography) were observed in 74 cases (61.2% of the transient hyperprolactinemia cases). On the contrary, in 25 (12.4%) of the 121 LPD cases who showed the hyper-luteinizing hormone (LH) syndrome (LH/FSH ratio > 1), only 9 (36%) had mature follicles. Of the remaining 55 cases who showed neither transient hyperprolactinemia nor the hyper-LH syndrome, 27 cases (49.1%) had mature follicles. Five of these patients indicated a significantly higher LH pulse amplitude despite depressed P<sub>4</sub> secretion in the luteal phase. From these results, it was concluded that the most common cause of LPD was transient hyperprolactinemia. The second cause of LPD was suspected to be disturbed follicle development due to the inappropriate ratio of LH/FSH in the hyper-LH syndrome. Another cause was speculated to be the primary failure of a response from the corpus luteum to LH. Treatments based on the conclusions mentioned above resulted in a 48.3% pregnancy success rate.