Abstract: The decision to accept a plea bargain is one of the highest stakes decisions under uncertainty that an individual can make. It is also an extremely important source of unwarranted disparity in the criminal justice system. This paper undertakes the most detailed empirical study to date of the plea bargain decision. We use a dataset of over 300,000 observations from 15 years of court cases across North Carolina. In order to overcome omitted variables bias, we make use of an instrumental variables strategy. In North Carolina a defendant seeking to minimize likelihood of incarceration is generally better off rejecting a plea, while one who wishes to minimize expected sentence length will take the plea. But underlying these general findings we find that the decision to plea or not is complex and varies substantially by race, type of crime, jurisdiction, attorney type, and judge. These findings have important implications both for optimal decision-making by defendants and for structure of the criminal justice system.