The study found that 20 percent of released suspects committed a pretrial new offense and 21 percent failed to appear in court. New offenses committed in the pretrial period were mostly petty misdemeanors or misdemeanors. The probability of committing a new offense was predicted by past felony and misdemeanor convictions, past or current bench warrants, if the instant offenses was drug- related or property-related, and offender age and employment status. The failure to appear seemed to be predicted by past or current bench warrants, past misdemeanor convictions, not having the option of voluntary surrender, and offender unemployment. The instrument developed as a result of this research has several advantages in predicting the risk of pretrial release. It has a large range of total scores, is not dependent on self-reported items, allows persons with prior convictions to be released only with conditions, provided a mechanism for shared responsibility, depends less on variables related to community ties, and seems to be more impartial to minority suspects. 24 notes, 26 references, and 3 appendixes