For more than two decades, advocates of malpractice system reform have claimed that the most damaging and costly result of the U.S. medical malpractice system is the practice of defensive medicine, in which physicians order tests and procedures primarily because of fear of malpractice liability. In this article, we discuss the issues raised by different definitions of defensive medicine and propose a working definition to guide measurement of the concept. We also consider the strengths and weaknesses of available approaches for measuring defensive medicine. Finally, we describe an empirical approach to measuring defensive medicine using clinical scenario surveys. The results suggest that, if physicians actually practice as they say they would in these surveys, defensive medicine does exist, although not to the extent suggested by anecdotal evidence or direct physician surveys. The results also suggest that defensive medicine varies considerably across clinical situations. In all of the scenarios, many physicians chose aggressive patient management styles even though conservative management was considered medically acceptable by the expert panels that developed the scenarios. In most cases, medical indications, not malpractice concerns, motivated clinical choices. Our results highlight the limitations of surveys as a method of measuring the extent of defensive medicine. The implications of managed care and health care reform for defensive medicine are also discussed.