SOCIAL WORKERS FACE MANY DIFFICULTIES IN WORKING WITH GROUPS OF DELINQUENTS WHOSE BEHAVIOR IS DISRUPTIVE AND IMPENETRABLE. THE STRUCTURED FORMAT OF A TASK-CENTERED SYSTEM HAS BEEN EFFECTIVE FOR A VARIETY OF DELINQUENT POPULATIONS. A CRUCIAL ELEMENT IN THIS APPROACH IS THE HIGH VALUE PLACED ON THE CLIENT'S AUTONOMY AND AVOIDANCE OF IMPOSED TREATMENT STRATEGIES. IN AN INITIAL INTERVIEW, THE SOCIAL WORKER EXPLAINS THE PURPOSES OF THE GROUP, EMPHASIZES THAT THE YOUTH MUST BE WILLING TO WORK ON BEHAVIOR PROBLEMS, AND ASSESSES THE COMMITMENT TO TREATMENT. THE SOCIAL WORKER FIRST TEACHES THE GROUP CONSENSUS DECISIONMAKING SKILLS, AND THE GROUP THEN DECIDES ON FORMAT AND GOALS. BECAUSE LOW SELF-ESTEEM IS A CRITICAL FACTOR IN DELINQUENT BEHAVIOR, THE SOCIAL WORKER MUST PROVIDE CONSTANT POSITIVE FEEDBACK CONCERNING THE STRENGTHS OF GROUP MEMBERS. THE WORKER MUST ALSO ELIMINATE DISRUPTIVE BEHAVIOR BY FOCUSING THE GROUP'S ATTENTION ON THE PROBLEM AND ENLISTING THEIR HELP IN RESOLVING IT. THE SOCIAL WORKER CONTRACTS TO MEET WEEKLY WITH INDIVIDUAL MEMBERS TO IDENTIFY PROBLEM AREAS, ASSESS THEIR IMPACT, AND ASSIST IN DEVELOPING REPLACEMENT BEHAVIORS FOR PARTICULAR PROBLEMS SUCH AS LOSING ONE'S TEMPER. TO TEST THIS APPROACH, TWO GROUPS OF DELINQUENTS INCARCERATED FOR NONSTATUS OFFENSES WERE SELECTED FROM A STATE CORRECTIONAL FACILITY. THE EXPERIMENTAL GROUP OF FIVE YOUTHS MET WEEKLY FOR 8 WEEKS WITH A SOCIAL WORKER EXPERIENCED IN THE TASK-CENTERED APPROACH, WHILE THE CONTROL GROUP OF EIGHT YOUTHS WAS INVOLVED IN UNSTRUCTURED RAP SESSIONS. MEASURED BY THE TENNESSEE SELF-CONCEPT SCALE, THE EXPERIMENTAL GROUP SHOWED SIGNIFICANT IMPROVEMENT IN SELF-ESTEEM LEVELS WHILE THE SCORES OF THE CONTROL GROUP DECLINED SLIGHTLY. RATINGS FROM THE INSTITUTIONAL STAFF SUPPORTED THESE FINDINGS. THE IMPLICATIONS OF THE STUDY ARE LIMITED BY THE SMALL SAMPLE AND THE INSTITUTIONAL SETTING, AND THE METHOD SHOULD BE TESTED ON RESIDENTAL TREATMENT PROGRAMS AND YOUTHS LIVING AT HOME. FOOTNOTES ARE PROVIDED. (MJM)