Econometric analysis of data from Sweden between 1976 and 1979 shows that arresting and jailing has a substantial impact on accidents that would otherwise lead to serious impairment. These factors can explain 40 to 50 percent of the variation in accident rates. The withdrawal of drivers' licenses, which occurs in Sweden to protect the public rather than to punish criminal actions, has as great or greater impact on accident rates. Alcohol consumption appears to be positively related to accidents, as are the distance driven and, in some cases, the vehicle mix. Rainfall appears almost unrelated to accident levels, in contrast to results from Great Britain. It is not known whether the threat of revocation of the license or the revocation itself reduces accidents. Use of license withdrawals as a penalty for drunk driving has several advantages, including its imposition of the most costs on the driver and relatively little costs on society. Data tables, thirteen references, and an appendix presenting two graphs are included.