Comparisons of preimplementation and postimplementation data indicated consistency in the distribution of scores, the classification of defendants into risk categories, and the failure to appear (FTA) rates for each group. Moreover, the new system recommended a greater number of defendants for release on recognizance while simultaneously decreasing their FTA rates slightly. The system decreased the portion of defendants categorized as moderate risk while slightly increasing their FTA rates. The system categorized the same proportion of defendants as high risk, while slightly decreasing their FTA rates. The results thus suggest that the new recommendation system performed as well as projected during its first year of implementation. The more accurate identification of defendants in the moderate risk category also allowed for a greater range of release options, including supervised and conditional release, designed to reduce their risk for FTA. Three research samples were used to evaluate the new release-on-recognizance recommendation system: (1) a postimplementation sample that was drawn from a cohort of defendants arrested between November 2003 and January 2004; (2) a 2001 pre-implementation sample that was drawn from a 3-month cohort of arrests made from January through March 2001; and (3) a 2002 preimplementation sample that was drawn from a cohort of defendants who were arrested in the third quarter of 2002. The consistency of the new recommendation system was measured by comparing the projections made from the 2001 and 2002 preimplementation samples with the post-implementation results. Comparisons were restricted to those defendants who had been released prior to disposition and were therefore at risk for FTA. Figures