Data from 13 districts that supported experimental pretrial service programs and reported bail decisions and release outcomes were analyzed. Five types of bail are described: release on personal recognizance, release on unsecured bond, deposit bond, surety bond, and collateral bond. A variety of factors, including the seriousness of the charged offense, the district where the bail was imposed, and the offender's criminal record, are shown to affect the level of bail. Bail is revealed to be somewhat lower for defendants who had lived at the same address for several years than for transients, for women than for men, and for defendants with college and high school degrees than for those without. Holding other factors constant, there was no correlation between bail amount and race, age, drug use, income, and prior history of bail jumping. Moreover, data indicate that the longer a defendant is at risk, the greater the probability of misconduct: the probability of misconduct is 10 percent for defendants who were on bail for 90 days, 14 percent for defendants on bail for 180 days, and 17 percent for defendants free for 270 days. Prior criminal record and drug use also increase the probability of misconduct. Additionally, males, non-Caucasians, and defendants considered less stable socially and economically are more likely than other defendants to commit some form of pretrial misconduct. Five tables and 13 references are included.