The reforms most often proposed have been (1) changes in the legal tests for insanity, (2) abolition of the insanity defense, and (3) adoption of an alternative verdict of 'guilty but mentally ill.' It is widely believed that the issues related to these reforms are largely legal and normative. However, questions about the empirical consequences lie at the center of most reform. Past reforms of the insanity defense have been based largely on untested assertions and polemics instead of on actual experience and experimentation resulting in clear data. Although empirical research cannot provide all the information needed about the subject, replacing dogma about what ought to be with some empirical data about what actually exists can help clarify some of the complexities, minimize the problems, and guide the necessary revisions in policy and practices. 143 footnotes.