The analysis of data from clinical trials often includes subgroup analyses, which are performed to examine the treatment effect within various sets of patients based on baseline and/or demographic variables. The goals of these analyses are to establish the consistency of the results across the subgroups and to identify important prognostic factors. The p-values for such analyses are usually presented without any adjustment for the multiple analyses: This approach has been criticized because of the possibility of misleading false positives. Conservative approaches have been proposed to resolve this problem; however, these approaches are usually so conservative that significant results are rarely observed after adjustment. Here an approximate technique for use when the variable of interest has a normal distribution is presented.