The authors of this article test hypotheses derived from Irwin's rabble management thesis. The analysis uses data from 47,592 interviews conducted with jailed adults in 30 U. S. cities as part of the Arrestee Drug Abuse Monitoring program. Clearly, homeless persons are overrepresented among those arrested and booked into local jails. Bivariate analysis support a fundamental assertion of the rabble management thesis: Homeless are jailed not because of their dangerousness but rather their offensiveness. Homeless arrestees are distinct from their domiciled counterparts in terms of sociodemographic characteristics, previous experiences with alcohol and drug treatment, mental health, criminal justice systems, and alcohol and drug use histories. In addition, homeless are less likely than domiciled arrestees to be jailed for felonies and violent crimes but more likely to be charged with maintenance and property crimes. Logistic regression models confirm these differences, even after other factors are controlled. A discussion of the policy implications of these findings follows.