A medical device is any item that treats or diagnoses a health condition whose action is not primarily chemical or biological. The main focus of this article is the design and evaluation of clinical studies to establish the safety and effectiveness of different kinds of medical devices. Among nondiagnostic devices, therapeutic and aesthetic devices, including implants, are considered. Among diagnostic devices,
in vitro
diagnostic tests and diagnostic imaging systems are considered. Given their increasing importance for precision medicine, molecular diagnostic tests are given special attention. Clinical studies of medical devices pose unique statistical challenges in design and analysis. Statistical issues for therapeutic devices include the placebo effect, sham controls, inability to perform blinded studies, noninferiority, survival analysis, repeated measures, and historical controls. Diagnostic devices pose a very diverse set of statistical challenges, with markedly different design and analysis considerations than therapeutic devices. Given their prevalence and applicability, adaptive designs and Bayesian approaches to medical device studies are also discussed.