Three questions relevant to insanity decisions were examined: (1) What informational cues are weighed most heavily in the attribution of criminal responsibility? (2) How do verdict forms influence these attributions? (3) How do individuals' beliefs about insanity and responsibility influence decision making? A total of 181 responded to vignettes portraying an act by a mentally disordered defendent. Psychiatric jargon was avoided, so that attributions were not a function of diagnostic terminology. It was found that, under the traditional scheme of not guilty by reason of insanity (NGRI) verus guilty, level of mental disorder (schizophenia verus personality disorder) was the primary determinant of insanity decisions. Also, insanity judgements were more likely to be made for acts performed without planful intentionality. Under the alternative scheme of NGRI verus guilty but mentally ill (GBMI) verus guilty, mental disorder still controlled NGRI verdicts; a bizarre act increased the likelihood of a GBMI over a guilty verdict; and the GBMI verdict option reduced markedly the proportion of psychotic defendants found NGRI and the proportion of personality disordered defendants found guilty. There were no significant differences between diagnostic groups in the likelihood of being found GBMI. Most subjects preferred to utilize the GBMI option as a compromise verdict even in the face of very severe mental illness. Implications for both social policy and future research are discussed. 5 tables and about 60 references. (Author abstract modified)