LONGITUDINAL DATA WERE OBTAINED FOR A COHORT OF MALES BORN IN PHILADELPHIA, PA., IN 1945 AND EVALUATED FOR THE PERIOD 1959-1974. THE DATA INCLUDE INFORMATION ABOUT OFFICIALLY RECORDED POLICE CONTACTS, SELF-REPORTS OF OFFENSES, AND DISPOSITIONAL OUTCOMES FROM THE JUVENILE AND ADULT JUSTICE SYSTEMS. TWO MODELS ARE DEVELOPED FOR MEASURING RESTRAINT, EACH OF WHICH CONTAINS TWO MAJOR TERMS, ONE TERM ESTIMATES ANNUAL AND AGE-SPECIFIC OFFENSE RATES FOR INDEX OFFENSES, AND THE SECOND TERM MEASURES THE PERFORMANCE OF THE JUSTICE SYSTEM. BY COMPUTING PROBABILITIES OF CONVICTION AND INCARCERATION AND THE LENGTH OF INCARCERATION TIME SERVED BY OFFENDERS, THE LEVEL OF EFFICIENCY ACHIEVED BY RESTRAINT THROUGH THE JUSTICE SYSTEM IS CALCULATED. THE PRODUCT OF THE TWO TERMS (OFFENSE RATE AND JUSTICE SYSTEM PERFORMANCE) AND AN ESTIMATE OF THE NUMBER OF OFFENDERS WHO COME IN CONTACT WITH THE JUSTICE SYSTEM ALLOWS A DETERMINATION OF HOW MANY OFFENSES ARE PREVENTED BY INCARCERATION. BASED ON COHORT DATA, IT IS ESTIMATED THAT ABOUT 2.4 INDEX OFFENSES WERE PREVENTED FOR EACH CHRONIC OFFENDER INCARCERATED FOR 1 YEAR. THE INCARCERATION OF SUCH AN OFFENDER FOR 2 YEARS PREVENTED 4.8 INDEX OFFENSES. OVERALL, THE EFFICIENCY OF THE COMBINED JUVENILE-ADULT JUSTICE SYSTEM IN PHILADELPHIA FOR THE COHORT WAS 0.029; ON THE AVERAGE, OFFENDERS WERE INCARCERATED FOR LESS THAN 3 PERCENT OF 1 YEAR (11 DAYS) FOR EACH INDEX OFFENSE THEY COMMITTED. THE SYSTEM WAS MOST EFFECTIVE AGAINST CHRONIC OFFENDERS, AND IT WAS LEAST EFFECTIVE IN THE CASE OF JUVENILES. PUBLIC POLICY IMPLICATIONS OF THE FINDINGS ARE DISCUSSED. SUPPORTING DATA ARE PROVIDED FOR OFFENDERS AND RECIDIVISTS AND FOR SEVEN OFFENSE CATEGORIES (HOMICIDE, FORCIBLE RAPE, ROBBERY, AGGRAVATED ASSAULT, BURGLARY, LARCENY, AND AUTOMOBILE THEFT). APPENDIXES CONTAIN A LIST OF DATA SOURCES USED IN THE EVALUATION, ADDITIONAL DATA, AND THE INTERVIEW SCHEDULE EMPLOYED TO ELICIT SELF-REPORTS OF ILLEGAL BEHAVIOR. AN INDEX AND A BIBLIOGRAPHY ARE INCLUDED. (DEP)