Data were collected on 2,764 cases scheduled for a pretrial during March, April, and May, 1978. The data included type and seriousness of offense, number of charges, type of defense counsel, judge at pretrial and disposition, mode of disposition, and sentence or sanctions imposed. Additionally, prior record information was collected from prosecutor files for drunk driving cases. Formal, semistructured interviews were conducted with such persons as the supervisors of the municipal units of the prosecutor's offices and public defenders and six municipal court judges. The interviews focused on such issues as modes of case disposition, judicial styles in plea bargaining, and the operations of arraignment court. In addition, 10 judges were observed for several hours at a time in pretrial sessions and other functions. A stepwise regression model was used to analyze data. The findings were compared with the data from a 1979 study of the New Haven lower court. Outcomes were more costly to convicted defendants in Columbus. Fines were substantial, incarceration was not infrequent, and in traffic cases drivers license were in jeopardy. Furthermore, courtroom actors behaved as if the outcome was important. For example, defense counsel stalled at pretrial hoping for a more sympathetic prosecutor or bargain on the day of trial. To explain why outcomes were more punishing in Columbus than in New Haven an overview of local political cultures of the two cities was conducted. It showed that New Haven was a criminal court system that reflected the particularistic values of ethnic, religious, political, and family association in its rendering of swift, substantive justice. Columbus, by contrast, was a system that reflected the universalistic values of professional competence and technical efficiency in its rendering of swift but formal justice through the mechanisms of an adversary system as applied to a misdemeanor court. The political culture of Columbus bred a climate of severity. Statistical data and about 20 references are included.