Multiple regression analysis and discriminant function analyses were applied to the question of prediction of therapeutic success or failure to clozapine therapy, using a non-biological predictor pool of 46 items. This effort was not successful, accounting for less than 25% of outcome variance under optimal conditions. However, it does appear that while clozapine efficacy in treatment-resistant schizophrenic patients in general is superior to such neuroleptics as chlorpromazine or haloperidol, the "paranoid" subgroup appears to benefit most.