The evaluation findings from Phase I show that the new system for recommending defendants for ROR performed as predicted. Compared to the old system, it recommended a substantially higher proportion of defendants for ROR and decreased the proportion of moderate-risk defendants. The new system showed a 2-percent increase in the overall ROR rate at arraignment compared to the old system. Under the old system, the low-risk and moderate-risk groups did not differ regarding release decisions at arraignment. Under the new system, defendants recommended for ROR had the highest ROR rate, which was followed by the moderate-risk group. Defendants who were not recommended for ROR by the Criminal Justice Agency (CJA) had the lowest ROR rate. CJA's new recommendation system is based on a defendant's ties to the community and criminal history. It considers whether the defendant has an address in the New York City area; has a working residential telephone or cell phone; is employed, in school, or in a training program full-time; expects someone at his/her arraignment; has any prior bench warrants; or has any open case counts. Data for Phase I of the postimplementation evaluation were drawn from a cohort of arrests made from November 1, 2003, through January 31, 2004. The dataset used in the analysis contained 59,551 defendants who were held for arraignment in criminal court. 9 tables