Abstract
Psychiatrists are called upon to make judgements on the future dangerousness of mentally disordered subjects in civil commitment procedures, in the criminal courts, and during the decision process on the release of offenders on indeterminate sentences or committals. The ability of psychiatrists to make these judgements is increasingly under challenge. The difficulties of making useful predictions when the base rate for the event to be predicted is low, is now well recognised. Less obvious are the problems attendant upon making socially useful predictions on psychiatric grounds in populations with a high base rate for future offending. The evidence pertaining to the level of violence amongst the mentally disordered is reviewed. The lesson to be drawn from the empirical evidence is that mental abnormality of and in itself contributes littie to the prediction of the predisposition to act violently. The question remains as to whether there are definable groups or classes within the generality of mentally abnormal individuals for whom there is an increased risk of future violence. It would be compatible with both the research studies and common clinical impression if the mentally abnormal contained subgroups with unusually violent predispositions balanced by larger groups with less than average propensities to aggression. The literature is largely inadequate to delineate such high risk groups with the degree of certainty ideally needed to instruct clinical decisions, but does suggest that such groups exist and are capable of further empirical definition.