This open-label, prospective, randomized, multicenter trial compared the incidence of amenorrhea in 54 postmenopausal women (mean age, 54.9 ± 0.6 years) who underwent six 4-week cycles of continuous hormone replacement therapy combining a progestin—nomegestrol acetate 2.5 mg/d—plus one of three estrogens: percutaneous 17β-estradiol gel (1.5 mg/d, group A), transdermal 17β-estradiol patch (50 μg/d, group B), or oral estradiol valerate (2 mg/d, group C). Based on an intent-to-treat analysis, the rate of amenorrhea varied significantly according to which estrogen preparation was used. Calculated cycle by cycle, rates of amenorrhea were 67% to 83% for group A, 25% to 56% for group B, and 53% to 61% for group C. Overall rates of persistent amenorrhea were not statistically different between groups for cycles 1 through 3, but for cycles 4 through 6, significantly more women in groups A and C (67% and 46%, respectively) experienced amenorrhea than did those in group B (12%). Amenorrhea rates for the entire six-cycle period were 78% for group A, 48% for group B, and 60% for group C. These differences were not statistically significant. The differences in rates could not be attributed to endometrial atrophy, since when measured by transvaginal sonography, endometrial thickness did not differ significantly between groups. Of the original population, 7% withdrew prematurely because of bleeding. The data for all three groups confirmed that in two out of three women, the occurrence of amenorrhea during the first three cycles predicted continuation of amenorrhea during subsequent cycles and that for 51% of women, ≤10 days of bleeding during the first three cycles predicted amenorrhea during the last three cycles. Calculated as a function of the number of women included in the trial, the percentage of amenorrheic women (evaluated cycle by cycle or for the second three-cycle period) was highest when the progestin was combined with percutaneous 17β-estradiol gel, although findings were similar with estradiol valerate. The percutaneous 17β-estradiol gel was also associated with a higher percentage of amenorrheal cycles than was estradiol valerate or transdermal estrogen, although differences were significant only in comparison with the transdermal formulation. This difference may have positive clinical implications.