Background Japanese road traffic law was amended in 2002 and 2007 to increase the penalties for drink-driving in response to media coverage, publicity campaigns, and debates following high-profile alcohol-related motor-vehicle crashes in 1999 and 2006.

Objective To test the hypothesis that the proportion of crashes involving drink-driving started to decline before the law amendments, because of changes in social norms and driver behaviour after the high-profile crashes.

Methods In order to assess the impact of the cases in 1999 and 2006, time-series analyses were used to examine the trends in the proportion of crashes involving drink-driving, and whether there were abrupt changes in the level or slope at the expected time points, using monthly police data for the period between January 1995 and December 2008.

Results In 1999, the proportion of alcohol-related fatal crashes in which the driver had a blood alcohol concentration (BAC) ≥0.5 mg/ml started to decline with a slope change of −0.09 percentage points per month (95% CI −0.15 to −0.03) but no level change, whereas there were no changes for drivers with a BAC <0.5. In 2006, the trends for drivers with a BAC ≥0.5 or <0.5 showed significant level declines of −3.1 (−5.0 to −1.2) and −1.7 (−2.5 to −0.9) percentage points, respectively, but no slope changes.

Conclusions Media coverage of high-profile crashes, and subsequent publicity campaigns and debates might have altered social norms and driver behaviour, reducing the proportion of alcohol-related crashes before the introduction of more severe penalties for drink-driving.