THIS PAPER DISCUSSES THE MANY TIMES PSYCHIATRISTS AND COURTS ARE ASKED TO PREDICT A PERSON'S FUTURE BEHAVIOR, THE SERIOUS CONSEQUENCES OF A DETERMINATION OF DANGEROUSNESS, THE LACK OF CLEAR CRITERIA FOR A DETERMINATION OF DANGEROUSNESS, AND SOME OF THE TECHNICAL PROBLEMS ASSOCIATED WITH THE PREDICTION OF DANGEROUS OR VIOLENT BEHAVIORS. IT IS POINTED OUT THAT SOME STUDIES HAVE FOUND THAT REPEATED ARRESTS FOR DRUNKEN DRIVING ARE MORE OFTEN ASSOCIATED WITH VIOLENT BEHAVIOR THAN IS COMMITMENT TO A MENTAL HOSPITAL. THE ACTUARIAL OR STATISTICAL APPROACHES TO PREDICTION ARE EXAMINED, AND THE PROBLEMS WHICH HAVE ARISEN BECAUSE SOCIAL CONTROL AND TREATMENT HAVE BEEN CONFOUNDED ARE REVIEWED. IT IS POINTED OUT THAT CONFINEMENT TO A MENTAL HOSPITAL IS USED OFTEN AS PUNISHMENT, WHILE SOME COURTS HAVE HELD THAT A FINDING OF 'NOT GUILTY BY REASON OF INSANITY' DOES NOT AUTOMATICALLY LEAD TO COMMITMENT OF AN INDIVIDUAL TO A MENTAL HOSPITAL. THE FOLLOWING PROBLEMS ARISE FROM THIS CONFUSION OF THE GOALS OF TREATMENT AND MAINTAINING SOCIAL ORDER: A LACK OF CLEAR POLICY REGARDING VIOLENT OFFENDERS, A FAILURE TO DISTINGUISH BETWEEN BEHAVIOR WHICH IS A NUISANCE TO OTHERS AND THAT WHICH IS TRULY DANGEROUS TO OTHERS, CONFUSION OVER LENGTH OF INCARCERATION AND TYPE OF INCARCERATION WHICH SHOULD BE SPECIFIED, AND A LACK OF GOOD RESEARCH AND SOUND DATA FOR MAKING THESE DECISIONS. THIS DISCUSSION CITES JUDICIAL DECISIONS AND RESEARCH PAPERS. AN EXTENSIVE BIBLIOGRAPHY IS PROVIDED. (GLR)