The purpose of this study was to examine predictors of service gaps in a sample of youth involved in the juvenile court system. The study sought to investigate differences in demographic variables, delinquent/criminal behavior variables, family system variables, and juvenile court system variables between service recipients and non-recipients among court-involved juvenile delinquent youth. It also sought to create a predictive model that would correctly identify youth at the greatest risk of not receiving services. The study employed an analysis of 244 youth, 182 (75 percent) service recipients and 62 (25 percent) non-recipients, randomly selected from a larger pool of 500 juvenile court-involved youth. The findings failed to provide empirical support for the hypothesis that different demographic variables, delinquent/criminal behavior variables, family system variables, and juvenile court system variables could distinguish service recipients and non-recipients from one another. Although the analysis suggested that, relative to service non-recipients, service recipients were younger; more likely to have run away from home; tended to enter the program through the children’s services intake site; and had an active children’s services case worker and were involved in the mental health system, parents and guardians of non-recipients reported their youths to be significantly less assaultive at 6 months. These findings are inconsistent with research in the field that has suggested that demographic, delinquent/criminal behavior, and family system variables are important predictors of recidivism. Given the high recidivism rates among juvenile offenders and the fact that a significant percentage of juvenile offenders will become adult offenders, identifying ways to prevent service gaps should be of the highest priority. Having a predictive model in place makes this type of identification possible. If youths at risk of receiving no services are identified early, any number of strategies to ensure service receipt can be implemented. 2 tables, 31 references