This research project has developed a predictive model for identifying high risk offenders who are clients of a juvenile court system. The most important variables in the efficiency of the prediction capability of the model were as follows: (1) whether the delinquent had delinquent siblings and/or significant others as associates, (2) whether the delinquent had school problems, (3) what type of offense the delinquent committed (i.e., misdemeanor or felony) and (4) the sex of the delinquent. The proposed model in this study is no panacea for the problems besetting the juvenile justice system but it could make a meaningful contribution in addressing one of the system's most serious problems, namely a reduction in the recidivism rate. (Author abstract)