The belief in the psychiatrist's ability to predict the likely dangerousness of a patient's future behavior is almost universally held, yet it lacks empirical support. This paper is an attempt to organize the problem of prediction of dangerousness. Recent empirical studies which begin to isolate the characteristics of danger as well as supporting data from the psychiatric, socilogic, and legal literature are detailed. The histories of a group of mentally ill criminals recently released from prison which I have followed over time, are examined in terms of possible determinants of their dangerousness, and predictions are made and compared to those made by others.
Indications are that social factors have primacy, while psychologic and biologic factors require further study.