Arrests and jailing of drunken drivers have reduced accidents involving serious injuries. These measures explain 40 to 50 percent of the variations in accidents involving serious injuries and those leading to serious impairment. The withdrawal of the driver's license, which is not regarded as a punishment in Sweden, may have as great or greater effect for reducing accidental injury. Arrests and accidents have a two-way causal pattern, according to the testing of a model which posits a simultaneous system. The tests used the records of all fatal and serious accidents in Sweden from 1976 through 1979 and all arrests and convictions for drunk driving over the same time period. The level of alcohol consumption, as indicated by the number of arrests for drunkenness, is a significant predictor of accidents, even when nonseasonalized data for drunkenness are used. The distance driven and the mix of vehicles are also positively related to accidents. Rainfall is not statistically significant in most formulations. The analysis does not permit a conclusion regarding whether deterrence, incapacitation, or both are operating to control accidents through the use of sanctions. Jail and license withdrawal are both valid options for the control of accidents due to drunk driving. However, the sancions affect fatal accidents differently than they do accidents involving serious injury. Figures, data tables, footnotes, a list of 13 references, and an appendix defining the study variables are supplied.