SUBJECTS FOR THIS STUDY WERE 536 BOYS BETWEEN 15 AND 18 YEARS OF AGE. SUBJECTS WERE RECRUITED FROM THREE CORRECTIONAL FACILITIES, A COMMUNITY-BASED JOB TRAINING PROGRAM, AND A LOCAL INNER CITY PUBLIC SCHOOL. COMPARISONS OF THE SELF-REPORTED AND COURT IDENTIFIED VIOLENT OFFENDERS WERE BASED UPON EXTENSIVE TESTS AND QUESTIONNAIRES YIELDING PSYCHOLOGICAL, SOCIOLOGICAL, DEMOGRAPHIC, FAMILY BACKGROUND AND INTERACTION, AND LEGAL DATA. THE DATA WERE ANALYZED USING A MULTIPLE REGRESSION ANALYSIS. THE FINDINGS OF THE STUDY REVEALED SEVERAL COMMONALITIES AMONG YOUTHS WHO WERE ARRESTED FOR COMMITTING A VIOLENT ACT AND THOSE WHO ADMITTED DOING SO BUT WERE NEVER CAUGHT. FOR BOTH GROUPS, THE MOST POWERFUL 'PREDICTORS' OF THEIR VIOLENCE WERE FAMILY-RELATED MEASURES SUCH AS LACK OF INVOLVEMENT IN CONSTRUCTIVE FAMILY FUNCTIONS, PARENTAL DEFIANCE, OR DISRUPTIVE FAMILY BEHAVIOR. SIGNIFICANT DIFFERENCES BETWEEN THE GROUPS WERE ALSO FOUND. APPREHENDED YOUTHS WERE CHARACTERIZED BY MORE ALIENATION FROM FAMILY, FUTURE FUTILITY, LOWER INTELLIGENCE, POORER PLANNING ABILITY, AND PROBLEMS WITH AUTHORITY THAN THE SELF-REPORT GROUP. IMPLICATIONS OF THESE FINDINGS ARE DISCUSSED. (AUTHOR ABSTRACT MODIFIED)