The purpose of this study is to show the potential of improved traffic safety from compulsory alco-lock for those convicted for drink-driving. About 800 of all drivers yearly involved in police reported road accidents with casualties in Sweden are suspected for driving under the influence of alcohol or other drugs. However, many of those drivers have never been convicted for drink-driving, many drivers drive without driving license and many drivers drive cars that are borrowed or stolen. Therefore it is vital to quantify the proportion of the 800 drivers involved in accidents that might be using an alco-lock if that would be compulsory for drivers convicted for drink-driving. One way to assess the potential of improved traffic safety of compulsory alco-lock is to assume that it is drivers with valid or cancelled driving licenses that are affected. The mean number of drivers involved in police reported road accidents with casualties, where the driver has a valid or cancelled driving license and is suspected of drunk driving, is 626 per year, with a 95 percent confidence interval ranging between 605 and 649 drivers. 17 percent of those drivers where convicted for drink-driving at least once during the 10 years preceding the accident and the 95 percent confidence interval is ranging between 16 and 19 percent. A regulation of compulsory alco-locks for drivers convicted of drink-driving is estimated to reduce the number police-reported road accidents with casualties with with 571 during the first ten years, of which 218 are accidents with fatalities or serious personal injuries.