Study subjects were persons arrested for DUI (driving under the influence) between January and June 1993, with administrative per se license suspension actions during this time period and subsequent convictions for DUI or reckless driving; a total of 53,217 offenders were used in the study. The dependent variable was a DUI event. Independent variables were a range of factors believed to relate to DUI recidivism. Statistical analyses predicted whether subjects would recidivate during the year following their arrest. Findings show that the blood alcohol content (BAC) level at arrest is statistically related to the probability of recidivism during the year following arrest. Further, high rates of recidivism at high BAC's suggest alcohol dependency, and at low BAC's other impairing substances are suggested. Prior 2-year total traffic convictions can increase predicted recidivism as much as a large increase in BAC level. The study concludes that the probability of DUI recidivism predicted by a sample model that uses BAC, prior 2-year total convictions, and offender level could be used by presentence investigators, judges, or in administrative settings to determine appropriate sanctions, treatment program assignment, or other remedial measures. Support for various DUI remedial policies and a suggestion for further research are also considered. For volume 2, see NCJ-167941. 29 tables, 13 figures, and 37 references