Because poor compliance introduces a major risk of bias in the interpretation of the results of a therapeutic trial, it is an important element to consider. At the planning stage, factors known to be associated with poor compliance should be recognized. The different methods of evaluating compliance, either clinical or biological, should be reviewed and the best strategy selected. During the therapeutic trial, the objective is to maintain an appropriate level of compliance. Patients, investigators, and sponsors have different options and responsibilities. The analysis should incorporate compliance as a specific variable in order to help test the robustness of the data. Compliance constitutes by itself a specific outcome measure. Compliance should be an integral part of study reports and publications, but it is frequently not discussed.