The effects of IUCD's on ovarian and menstrual function were studied in 10 women throughout at least 2 menstrual cycles, one in which an IUCD was in place and the other was a control. Ovarian function was monitored and ovulation was dated by serial measurements of urinary estrogens and pregnanediol and of plasma progesterone. The steroid levels in all cycles showed the characteristic pattern of the ovulatory cycle with the values falling within the normal ranges. No significant differences in output of estrogens or pregnanediol could be detected between control and IUCD cycles and there was no evidence that the IUCD influenced follicular development, ovulation or the function of the corpus luteum. However, the lengths of the luteal phases, calculated as the interval between the mid-cycle estrogen peak and onset of menstrual bleeding, were shorter in the IUCD cycles than in their controls, the mean differences being 2.3 days which was highly significant (p < 0.001). Even so, the durations of the luteal phases in 70% of the IUCD cycles were still within the range found in normal fertile women and therefore the shortening could not itself explain the contraceptive action of the IUCD. The estrogen and pregnanediol values were significantly higher on the day before onset of bleeding in the IUCD cycles than in their controls. It was concluded that the shorter luteal phases in the IUCD cycles were not due to earlier hormone changes but were due to local effects of the IUCD which caused earlier shedding of the endometrium. (/ Clin Endocrinol Metab 36: 1125, 1973