To assess the performance of the new recommendation system for ROR, it was applied to the most recent validation sample of defendants (arrests in 2001). The new system was compared with the prior system regarding the distribution of defendants in various risk categories and their corresponding rates of failure to appear (FTA) in Criminal Court. The new recommendation system was found to have several advantages over the old system. First, the new system takes into account a defendant's criminal history as well as community ties; the previous system had assessed a defendant's risk of flight based only on community ties. Second, the new system is based on research conducted on a citywide sample of defendants; whereas, the old system was based only on research conducted with a sample of Brooklyn defendants. Third, the new system predicts risk of flight for both Criminal Court and Supreme Court; the prior system predicted risk of flight only for Criminal Court appearances. Fourth, the new system places defendants into one of three categories based on their relative risk of flight: low risks have a score on a point scale that assigns them a much lower FTA rate; moderate-risk defendants have a greater risk of FTA than low-risk defendants; and high-risk defendants have a greater risk of FTA than moderate-risk defendants. The previous system barely distinguished between low-risk and moderate-risk defendants in the validation samples. Fifth, the new system improves on the previous one by significantly increasing the number of defendants classified as low risk, without any increase in FTA rates, thus reducing the pretrial jail population. Finally, by identifying defendants' level of risk for FTA, the new recommendation system allows for consideration of a variety of release options designed to reduce defendants' risk of FTA. 8 figures