The author reviews studies that address the question, "Are the mentally ill
dangerous?" She points out that as psychiatrists, we have the responsibili
ty of evaluating the mentally ill and making judgments about their dangerou
sness that may restrict their civil liberties. Therefore, the more practica
l question for us is: "Which mentally ill, under what circumstances, are da
ngerous?" She discusses data from her research group and others that show t
hat short-term predictions of violence can be relatively accurate, that we
are better at predicting violence for some patients than for others, that s
pecific symptom patterns in the acute phase of illness are related to viole
nt acts, that the most likely victims of violence by decompensating psychia
tric patients are caretakers rather than strangers, and that a history of v
iolence, co-morbid substance abuse, and treatment noncompliance are related
to a higher risk of violence in psychiatric patients.