THE NEED FOR THE CAPABILITY TO PREDICT IS CLEAR FOR ALMOST EVERY STEP OF THE WAY THROUGH THE CRIMINAL JUSTICE SYSTEM AND PARTICULARLY THE CORRECTIONS SEGMENT OF IT. DECISIONMAKERS ARE CALLED UPON TO DETERMINE WHO REQUIRES TIGHTER CONTROLS, CLOSER SUPERVISION, OR TOTAL REMOVAL FROM SOCIETY. TECHNIQUES SEEN AS HELPFUL IN MAKING THESE PREDICTIONS--THE CALIFORNIA BASE EXPECTANCY SCALE, THE NORM HOLT STATISTICAL DEVICE, AND OTHER RELATED TECHNIQUES--ARE REVIEWED, AND THEIR GENERAL LACK OF SUCCESSFUL APPLICATION IS NOTED. SERIOUS CONCERN IS EXPRESSED ABOUT THE STRONG TENDENCY TO OVERPREDICT LEADING TO THE IMPOSITION OF UNNECESSARY CONTROLS ONTO LARGE NUMBERS OF PEOPLE. SIMILARLY THE STATISTICAL PROBLEM OF PREDICTING THE RARE EVENT IS EXAMINED, WITH THE FINDING THAT THE USE OF EVEN SEEMINGLY ACCURATE RATING SCALES OR PROCEDURES LEAD TO HIGH LEVELS OF MISCLASSIFICATION TO THE DETERIMENT OF THE NONVIOLENT. AN ALTERNATE APPROACH IS RECOMMENDED. IN ESSENCE, IT INVOLVES RESHAPING ALL PREDICTIONS TOWARD DETERMINING WHICH INDIVIDUALS COULD MANAGE SATISFACTORILY WITH FEWER CONTROLS, WITH LESS SUPERVISION, OR FREE FROM ALL RESTRICTIONS. SUCH A PROCEDURE WOULD FUNCTION WITH A GREATER PERCENTAGE OF ACCURACY AND NOT IMPOSE ONTO FALSE POSITIVES IN THE SYSTEM UNJUST RESTRICTIVE MEASURES. THIS TYPE OF APPROACH MIGHT BE ADOPTED BY CORRECTIONAL RESEARCHERS, PLANNERS, AND THOSE IN CHARGE OF CLASSIFICATION SYSTEMS AND CODES OF ETHICS. DISCUSSION QUESTIONS AND NOTES ARE INCLUDED. (AUTHOR ABSTRACT MODIFED--DAG)