STUDIES ARE CITED TO SHOW THAT BEHAVIORAL SCIENTISTS ARE UNABLE TO PREDICT VIOLENT BEHAVIOR, EVEN AT MINIUM LEVEL OF ACCURACY. IT IS NOTED THAT MOST OFTEN PREDICTIONS OF VIOLENCE ARE EXCESSIVE. IT IS GENERALLY RECOMMENDED, THEREFORE, THAT DECISIONMAKING BASED ON THE PREDICTION OF BEHAVIOR SHOULD BE REDUCED IN FAVOR OF JUDICIAL DECISIONMAKING ORIENTED TOWARD DEVIANT ACTS ALREADY COMMITTED, WITH PRECISE SENTENCING ISSUED WHERE GUILT HAS BEEN DETERMINED BY DUE PROCESS UNDER THE LAW. IN THE CASE OF PRETRIAL PREVENTIVE DETENTION, IT IS SUGGESTED THAT EVIDENCE OF SERIOUS VIOLENT BEHAVIOR IN THE PAST BE WEIGHED MORE HEAVILY THAN PSYCHIATRIC TESTING IN MAKING DECISIONS ABOUT PRETRIAL DETENTION. REGARDING CIVIL COMMITMENT, IT IS HELD THAT THE LACK OF ABILITY TO PREDICT DANGEROUSNESS, COMBINED WITH THE SIMILAR BASE-RATES FOR VIOLENCE AMONG THE PSYCHOLOGICALLY DISTURBED AND 'NORMALS', INDICATES THERE IS NO EMPIRICAL BASIS FOR THE PREVENTIVE DETENTION OF PSYCHOLOGICALLY DISTURBED PERSONS WHO HAVE NOT COMMITTED A VIOLENT ACT. IT IS ALSO ARGUED THAT THERE IS NO EMPIRICAL OR MORAL JUSTIFICATION FOR CONTINUED EXPERIMENTATION WITH INDETERMINATE SENTENCING, SINCE THERE IS NO SOUND BASIS FOR PREDICTING BEHAVIORAL CHANGE. EARLY INTERVENTION TO EITHER RESTRAIN OR FORCIBLY TREAT THOSE PREDICTED TO BE PRONE TO VIOLENCE IS DEEMED A STIGMATIZING LABELING PROCESS THAT INTRUDES INTO PERSONAL FREEDOM WITH NO EMPIRICAL BASIS FOR SUCH ACTION. PRECISE LEGAL SAFEGUARDS ARE RECOMMENDED WHERE EARLY INTERVENTION IS PRACTICED. (RCB)