A method was developed to determine sample sizes for replication studies when the size of the effect being studied is uncertain. The method uses information from the original experiment to estimate a range of possible effect sizes. The sample size needed for a replication study is then calculated to ensure a desired level of statistical power across this range of effect sizes. A FORTRAN 77 program was created to quickly calculate the sample size needed for replication studies that involve comparing means, correlation coefficients, or proportions.