EIGHTY-SIX FAMILIES OF DELINQUENTS WERE RANDOMLY ASSIGNED TO ONE OF FOUR TREATMENT CONDITIONS: NO TREATMENT CONTROLS, A CLIENT-CENTERED FAMILY APPROACH, AN ECLECTIC-DYNAMIC APPROACH, AND A BEHAVIORALLY ORIENTED SHORT-TERM FAMILY SYSTEMS APPROACH. THE SHORT-TERM FAMILY SYSTEMS APPROACH INVOLVED A SET OF CLEARLY DEFINED THERAPIST INTERVENTIONS FOCUSED ON CHANGING FAMILY INTERACTION PATTERNS IN THE DIRECTION OF GREATER CLARITY AND PRECISION OF COMMUNICATION, INCREASED RECIPROCITY OF COMMUNICATION AND SOCIAL REINFORCEMENT, AND CONTINGENCY CONTRACTING EMPHASIZING EQUIVALENCE OF RIGHTS AND RESPONSIBILITIES FOR ALL FAMILY MEMBERS. A PROGRAM EVALUATION MODEL WAS DEVELOPED TO MEASURE THREE LEVELS OF PREVENTION EFFECTS: PROCESS CHANGES IN FAMILY INTERACTION AT THE TERMINATION OF TREATMENT (TERTIARY PREVENTION); RECIDIVISM RATES OF IDENTIFIED DELINQUENTS 6-18 MONTHS FOLLOWING TREATMENT (SECONDARY PREVENTION); AND RATE OF SIBLING CONTACT WITH THE COURT 2.5-3.5 YEARS FOLLOWING INTERVENTION (PRIMARY PREVENTION). THE FAMILY SYSTEMS APPROACH, WHEN COMPARED TO THE OTHER CONDITIONS, PRODUCED SIGNIFICANT IMPROVEMENTS IN PROCESS MEASURES AND A SIGNIFICANT REDUCTION IN RECIDIVISM. TO EVALUATE THE PRIMARY PREVENTION IMPACT, JUVENILE COURT RECORDS WERE EXAMINED TO OBTAIN RATES OF COURT REFERRALS FOR SIBLINGS OF INITIALLY REFERRED DELINQUENTS. RESULTS INDICATED THAT ONLY 20% OF FAMILIES IN THE TREATMENT CONDITION HAD SUBSEQUENT COURT CONTACTS FOR SIBLINGS, COMPARED TO A 40% RATE FOR NO-TREATMENT CONTROLS AND 59% AND 63% RATES FOR THE ALTERNATIVE TREATMENT CONDITIONS. REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED--RCB)