Beginning in 1961, various jurisdictions began allowing defendants to be released on their own recognizance while awaiting their trial. However, as this practice expanded to include almost all State and Federal jurisdictions, concerns emerged about public safety. In response, the first Federal preventive detention statute was passed allowing jurisdictions to consider safety risk and risk of flight when making pretrial release decisions. Currently, the New York State Criminal Procedure Law (CPL) does not explicitly permit the consideration of "dangerousness" when deciding pretrial release conditions. Given the fact that knowledge about pretrial recidivism among released defendants is limited and inconclusive, it is instructive to examine the rate at which these defendants commit new crimes while awaiting trial. The likelihood of pretrial recidivism was first examined by the New York City Justice Agency, Inc. (CJA) in 1989 using a random sample of defendants. In order to more fully examine this issue, the CJA decided to expand the study to a group of defendants arrested in 2001. The study examined whether defendants recommitted the same type of offenses, whether there were consistent predictors of rearrest across diverse samples, and whether the same variables predicted both pretrial rearrest and risk of flight. The analysis was performed on a database of 67,848 criminal defendants who were arrested between January 1, 2002, and March 31, 2001, in New York. The dependent variable was pretrial rearrest, which was used as a measure of recidivism. Numerous independent variables were examined for their effect on pretrial rearrest including community cohesion, criminal history, type and severity of the top arrest charge, demographic attributes, and case processing characteristics. Results of logistic regression analysis indicated that half of the defendants initially arrested for felonies were rearrested for felonies. Several variables were discovered to be consistent predictors of pretrial rearrest in both the 1998 and 2001 samples. These predictive variables include having a criminal history, being Black, being a younger defendant, being uneducated and unemployed, having a telephone, and being released on own recognizance while awaiting trial. Finally, the variables predictive of pretrial rearrest were not the same as the variables predictive of risk of flight. Tables, references, appendix