The sample consisted of all juveniles who were brought to family court for a second time on delinquency charges in two urban New Jersey counties between September 1993 and April 1994. The following 12 items were used to form a prospective risk instrument: early age of court docketing, lack of parental supervision, criminality in the family, parental alcohol/drug use, poor school performance, school behavior problems, negative peer influence, neurological dysfunction, past physical abuse, lack of impulse control, substance abuse, and early onset of behavior problems. A seven-item "retrospective" risk instrument was developed by using recidivism data. The instrument is retrospective in the sense that recidivism criteria (over an 18- month period) were known and used as dependent variables in the regression analysis that "predicted" recidivism that had already occurred. The findings show that the vast majority of all chronic offenders can be identified by screening second-time offenders with a risk instrument. The authors recommend that such risk instruments be used to focus benign intervention on high-risk, second-time juvenile offenders. 4 tables, 3 figures, and 49 references