Objective: To compare the effectiveness and acceptability of three regimens of postcoital contraception., Design: Randomised group comparison of ethinyloestradiol 100 micrograms plus levonorgestrel 500 micrograms repeated after 12 hours (Yuzpe method); danazol 600 mg repeated after 12 hours; and mifepristone 600 mg single dose., Setting: Community family planning clinic., Subjects: 616 consecutive women with regular cycles aged 16 to 45 years., Main Outcome Measures: Number of pregnancies, incidence of side effects, and timing of next period., Results: The raw pregnancy rates (with 95% confidence intervals) for the Yuzpe, danazol, and mifepristone groups were 2.62% (0.86% to 6.00%), 4.66% (2.15% to 8.67%), and 0% (0% to 1.87%) respectively. Overall, these rates differed significantly (chi 2 = 8.988, df = 2; p = 0.011). The differences between the mifepristone and Yuzpe groups and between the mifepristone and danazol groups were also significant. Side effects were more common and more severe in the Yuzpe group (133 women (70%)) than in either the danazol group (58 (30%)) or the mifepristone group (72 (37%)). The Yuzpe regimen tended to induce bleeding early but mifepristone prolonged the cycle. Three women bled more than seven days late in the Yuzpe group compared with 49 in the mifepristone group., Conclusions: Mifepristone was effective in reducing expected pregnancy rates and the Yuzpe method also had a clinical effect. Danazol had little or no effect. A further multicentre trial is needed.