PSA's release recommendations changed dramatically after introduction of the new method. Recommendations for unrestricted personal recognizance (PR) release increased sharply for both felonies and misdemeanors. During the first 18 months of the new system, an average of l3 percent of felony cases and 23 percent of misdemeanor cases received unrestricted PR release recommendations, as compared with fewer than 1 percent of the cases under the old system. Under the new method, PSA eliminated the 'no recommendation' category and added a category of 'other' recommendations, such as making an inquiry in open court to resolve conflicting information about the defendant's identity or address. Changes in judges' release decisions paralleled the changes in PSA's recommendations in one major respect: unrestricted PR release increased. Also, judges set fewer conditions for the average defendant under the new system. Failure-to-appear and pretrial arrest rates remained virtually the same. Thus, the less restrictive release practices did not result in increases in rates of pretrial misconduct. The study of whether risk assessments might be improved by using a quantitative forecasting model indicates that the approach has much merit and should be considered. Forty figures and eight tables are provided. See NCJ-95377 for appendixes.