Trials that compare drug and behavior therapies or evaluate combination therapy raise special methodological issues. This article reviews these methodological issues and, where possible, offers guidelines for addressing them. Sources of bias in the selection and recruitment of participants and in the measurement of treatment outcomes are discussed. In addition, methodological problems presented by the differing structures of behavior and drug therapy, by confounding variables, such as allegiance effects, differential expectations and preferences for drug or behavior therapy, and differential adherence with drug or behavior therapy also are reviewed. Issues in the selection of appropriate control groups are also discussed.