Studies of patient safety and comparative effectiveness entail unique methodologic challenges. These studies may be susceptible to systematic error, including selection bias, exposure misclassification, and outcome misclassification. They may also be vulnerable to random error, or confounding by a variable such as another drug, a disease, or the drug indication itself. Finally, special logistical issues can arise, including data access problems, difficulties in conveying the need for studies of certain interventions, and obstacles to gaining institutional review board approval. This article provides a conceptual overview of these methodologic issues.