On I March 1998, the Danish per se limit was lowered from 0.08 to 0.05% blood alcohol concentration (BAC) for motor vehicle drivers. Based on accident data and drivers' drinking habits before and after the amendment, the effect of the new limit has been evaluated. Interviews revealed a significant decrease in the number of drinks that drivers allow themselves to drink within a 2-h period before driving. The proportion of drivers, who would not drink at all or only have one drink, increased from 71% before the amendment to 80% after the amendment. Drivers with changed drinking habits most often stated the lower limit as the main reason for having less alcohol. However, based on accident data from the first year after the amendment, this has not resulted in a marked decrease in the proportion of injury accidents with impaired motor vehicle drivers (BAC greater than or equal to 0.05%) compared to all injury accidents. On the contrary, the proportion of fatal accidents with drink-drivers compared to all fatal accidents has increased in the after-period. The total number of drink-driving sentences were a little larger in 1999 than in 1997 because of the lower limit, but a significant change from higher towards lower alcohol levels can be seen. (C) 2002 Elsevier Science Ltd. All rights reserved.