This study investigates the short-term mortality effects of two age-based restrictions on legal access to alcohol in Germany. We exploit sharp differences in legal access to alcohol at 16 and 18 years by implementing a regression discontinuity design. We find discontinuous increases in deaths at both age cutoffs, which are mainly driven by a "novice driver effect", whereas legal access to alcohol plays a marginal role at most. Overall, our results indicate that a stepwise introduction to alcohol has, at most, a minor impact on drunk driving and mortality at age 16 and 18 years. This study thus provides fresh impetus to the ongoing debate on the "optimal" MLDA legislation.