Using the revised instrument, 30.5 percent of the cases were assigned to minimum supervision compared to only 7.3 percent of these cases when the original instrument was used. The failure rate under both the validated and original instrument remained virtually the same at 11.6 percent and 11.8 percent, respectively. Overall, the revised instrument distributes cases so they are more manageable, which allows a greater concentration of fixed resources on high-risk offenders under intensive supervision. Study data were obtained from three sources: a database that contained the risk and needs scores for probationers and parolees classified with the original instrument, probation and parole files closed between 1991 and 1996, and National Crime Information Center arrest records. A total of 2,097 cases were randomly divided into 2 equal subsamples, 1 for the construction of the instrument and 1 for validation. The construction sample was used to assess the performance of the original instrument. The performance of the revised instrument was tested with the validation sample to verify that the modified instrument improved the accuracy of classification. The best-fitting multivariate model was used to select items and set weights for the revised instrument. New cut points for classification levels were determined by examining the differentiation of base failure rates within groups. The implementation of the new instrument is described. 11 tables and 22 references