This study examines the predictive validity of Risk Management Systems (RMS), a fifth-generation risk/needs assessment. In contrast to traditional risk assessments, the RMS uses exemplar-based modeling to classify offenders as high or low risk. Using a sample (N = 830) of offenders on supervised release by the U.S. Probation Office, the current study explores the relationship between the RMS scores and recidivism. The results indicate that RMS recidivism and violence scores are predictive of arrests, technical violations, and unsuccessful terminations from supervision. However, only a limited number of offenders were assigned to the high-risk category, which may limit the utility of the instrument. Policy implications are discussed. (Published Abstract)