Findings showed that a number of significant risk and need variables (education, employment, and relapse prevention) had been previously overlooked by current risk assessments. This study provides an alternative risk assessment and classification scheme by calculating predictive probabilities of a client’s termination and recidivism through the use of logistic regression models. The research was composed of a two-phase study: the effect of risk and need factors on both a client’s termination from the DRC program and a client’s recidivism after release. The analysis of program termination identified four statistically significant variables that influenced clients’ termination from the DRC: two static risk variables (type of offense, and education) and two dynamic need variables (length of stay and relapse prevention). Three statistically significant variables were identified that influenced clients’ recidivism: two static variables (age on admission and risk points, or criminal history), and one dynamic variable (employment). Seven variables were used to create an appropriate risk/needs index in the prediction of termination from the DRC or recidivism after release. Data were collected through interviews and surveys of 273 clients on the period of their release; all participants were accused or convicted of nonviolent offenses. Tables, figures, appendix, notes, and references