Compliance with medication and medical appointments is presumed to have a critical influence on outcomes of medical interventions. However, compliance may not always be easily defined or accurately measured. No single method of measuring compliance with appointments or medication is applicable in all settings. The apparent effects of compliance on outcome may be both direct and indirect; research is needed to define more clearly how these effects may be mediated. Identifying effective methods of enhancing compliance requires accurate methods of measuring compliance. In addition, the effectiveness of interventions may decay over time, and differences in effectiveness in acute versus long-term settings have not been well studied. Researchers conducting clinical trials of medical interventions must evaluate compliance in the population studied and consider the potential impact of noncompliance on trial results and their generalizability. In some trial designs, data may best be analyzed by considering compliance a dependent or an outcome variable. Under appropriate circumstances, compliance may be considered an independent variable. Readers of the medical literature should consider how compliance was measured and analyzed when interpreting the results of clinical trials. Table 4 suggests criteria for critical appraisal of compliance-related issues in reports of clinical trials.