The sites were chosen because of their diversity in size, urban-rural and suburban character, volume of arrestees processed, administrative convenience, and quality of local records. Random samples of pretrial and sentenced inmates were drawn over 12 months (1982). Data collected on inmates included background information, current offense, prior record, confinement conditions, disciplinary problems, time and method of release, and disposition. The sources of referral to the jails varied by county and depended on the socioeconomic and geographic context as well as local criminal justice policies. In all sites, 48 to 66 percent of the inmates were charged with relatively minor violations of public order, violations of the court process, drunk driving, and traffic offenses. This was the first jail contact for most detainees. The pretrial admission population was predominantly male, youthful, uneducated, black or Hispanic, and unemployed or working at a lower class occupation; they had little or no cash when booked. A significant proportion had either a special admissions problem or were intoxicated. Most defendants held in the Yolo and San Francisco jails were released within 3 days after booking; those in Los Angeles were held longer. A significant proportion had their charges dismissed at court. Inmates sentenced to jail differed from pretrial inmates on one characteristic, denial of pretrial release while awaiting case disposition. The median length of time served for sentenced inmates varied dramatically across the three jails, but only a small percentage served more than 90 days. Disciplinary actions were rare. Policy implications of the study's results are discussed. The summary includes five references.