Many important drug properties cannot be predicted from in vitro or animal studies but must be quantified from studies of clinically relevant endpoints in humans. Clinical study designs include the randomized controlled trial (RCT) and a variety of observational designs. In RCTs, randomization usually ensures treatment group comparability with respect to other factors, so outcome differences reflect treatment differences per se. Properly conducted RCTs thus are the strongest study design, the mainstay of mandatory premarketing studies, and essential for evaluation of therapeutic efficacy. RCT limitations include frequent use of surrogate endpoints, limited power, short-term follow-up, and cross-contamination of study groups. For ethical reasons, some questions cannot be studied with RCTs. The simplest observational design, the case series, has limited value because it lacks a denominator. Limitations of cohort and case-control studies include misclassification, selection bias, and confounding. Despite their limitations, properly conducted experimental and observational clinical studies provide essential data for clinical practice.