Deficiency in the luteal phase has been shown during stimulated cycles using a protocol involving a GnRH agonist. The authors undertook a randomised prospective trial of supplementation by progesterone of the luteal phase and of early pregnancy in two hundred and seventy two patients requiring fertilisation in vitro (FIV), gamete inter-fallopian transfer (GIFT) or zygote inter-fallopian transfer (ZIFT). Either progesterone in solution in oil (50 mg/day) administered by intramuscular injection or micronized progesterone administered intra-vaginally (600 mg/d) were used as support for the luteal phase. Administration of progesterone in association with estradiol valerate was started on the day prior to oocyte puncture and was continued until the 12th week of pregnancy. The implantation rate was very close to the threshold of significance (P = 0.07) in favour of the patients given vaginal progesterone. There was a higher rate of clinical pregnancies (33.6 versus 26.7 p. cent) in the latter group, though this was not significant. While plasma progesterone (Pg) levels were lower in patients using vaginal progesterone, the abortion rate during the first three months was lower in this group (P < 0.05). Micronized progesterone administered vaginally was well tolerated by all patients. During stimulated cycles, notably by GnRHa, it thus proved to be more effective than Pg administered by intramuscular injection with regard to implantation and abortion rates.