AbstractInterim analyses are those that occur before the scheduled completion of a clinical study. The motivation for such analyses may be to see whether conclusive evidence is available concerning the aims of the study, or it may be simple curiosity. Statisticians disagree about the impact that interim analyses have on inferences that can be drawn from the study. The significance testing view insists that conclusions from a study with interim analyses are different than from one without—even though the data are identical. In the Bayesian view there is no penalty for interim analyses: study results can even be monitored continually without changing the conclusions. Both views are explained and recommendations for designing and analyzing studies with interim analyses are made.