In the field of diabetes, pharmacoepidemiologic (PE) studies are numerous and represent an essential tool to assess the real life effects of diabetes drugs. However, the specificity of the study of drugs used in diabetes, characterized by a high frequency of switches, interruptions and add-on, must be kept in mind by pharmacoepidemiologists in order to perform well-conducted studies and avoid several biases. In this article, the authors discussed the specific interest of PE studies in the field of diabetes, provide a brief overview of biases that can affect those studies, and suggest pragmatic solutions for more correct results and appropriate interpretation.