The Insanity Defense Reform Act (IDRA) of 1984, passed by the United States Congress after two years of hearings, illustrates the dangers of making law in the absence of evidence. In attempting to produce fewer not guilty by reason of insanity (NGRI) verdicts, the lawmakers changed the test definition of insanity, banned ultimate opinion testimony by experts and altered the burden and standard of proof. Yet, when each of these changes was empirically tested, it failed to produce fewer NGRI verdicts. From an empirical vantage point, research now is exploring the common sense constructs of sane and insane that mock jurors consistently appear to use, and these turn out to be complex, relevant and powerfully determinative of verdict. Evidence for partial responsibility judgments, along with two types of culpability judgments--at the moment of the act for the act, and for earlier actions that may bring about the mental disability--have been found. Finally, a sequential schema for insanity, which assesses both types of culpability and their degree has been tested, and the results show the least variance and the greatest coherence with other judgments and ratings of responsibility.