The standard method for estimating the effectiveness of emergency contraceptive pills (ECPs) uses external data to calculate the proportion of expected pregnancies averted by the treatment. Because these data may not be applicable to ECP study populations, this approach could result in substantial overestimation of effectiveness. We used data from two published randomized trials of the levonorgestrel and Yuzpe ECP regimens to calculate the minimum effectiveness of the levonorgestrel regimen. Conservatively assuming that the Yuzpe regimen was entirely ineffective in these trials, we estimate that the levonorgestrel regimen prevented at least 49% of expected pregnancies (95% confidence interval: 17%, 69%). Because physiologic data suggests that the Yuzpe regimen does, in fact, have some efficacy, the effectiveness of the levonorgestrel regimen is likely to be higher than our minimum estimate.