One hundred twenty participants functioned as mock-jurors and as members of deliberating juries in an experiment designed to assess the impact of dispositional instruction on verdicts rendered in an insanity trial. Consistent with prior research (K. E. Whittemore & J. R. Ogloff, 1995), dispositional instruction had no effect on the verdict preferences of individual jurors prior to deliberating. Yet, as expected, the instruction manipulation had a major impact on postdeliberative decisions (i.e., group verdicts; individual juror verdict preferences). Content analyses of jury deliberations revealed that postdeliberative shifts toward harsh verdicts in uninstructed juries and toward lenient verdicts in instructed juries were mediated by the impact of the Instruction manipulation on the content of jury deliberations: uninstructed juries feared that an acquitted-insane defendant would be freed to act again, whereas instructed juries recognized that finding for an insane defendant implied his retention and treatment. Implications of these results for both legal policy and the conduct of mock-trial research are discussed.