Comparisons of the Stockton trend in nighttime accidents with four preselected control cities and with the nighttime fatal crash rate in California and the United States as a whole demonstrated that the Stockton accident experience for the hour 8 p.m. to 4 a.m. was significantly different from any of the comparison accident trends. Time series analysis indicates there was a significant reduction in nighttime crashes, both on weekends and on weeknights in Stockton. This reduction was 10-15 percent over the project period. No similar reduction was found in the daytime accidents in Stockton or in the nighttime accidents in comparison cities. Evidence for the impact of the special drunk-driving patrol was strengthened by the results of roadside surveys conducted for a 3-month baseline period before the project began and then for the period of the project's operation. 4 figures, 4 tables, 11 references.