Since 1971, the school has used the 'Phoenix model' of reeducating rather than punishing offenders and has aimed to provide information on the consequences of drinking and driving. Random samples of 520 former students and 120 recidivists were obtained so that the sociodemographic variables of first offenders and recidivists could be compared. A subset of 30 former students was randomly selected for comparison to 30 randomly selected offenders who had not been sent to the school. All subjects were convicted between January 1, 1970, and December 31, 1973. Followup began from the date of school assignment or fine and ended 36 months later. The results did not indicate statistically significant differences between the control and treatment groups, although subjects in the treatment group had higher prior rates of traffic violations, points, and accidents than did controls. While the treatment group had about one-fourth as many subsequent DWI convictions, the difference was not statistically significant. The analysis of differences between recidivists and first offenders showed that recidivists were of higher occupational status, had a more serious alcohol problem, and were more likely to be involved in an accident at the time of arrest. These results suggest that the school might be more effective if first offenders and recidivists were treated separately and if language, content, and techniques used in the training were lailored to each group's level of understanding and environmental circumstances. Data tables, footnotes and an eight-item reference list are included.