We exploit arguably exogenous train schedule changes in Washington DC to investigate the relationship between public transportation provision, the risky decision to consume alcohol, and the criminal decision to engage in alcohol–impaired driving. Using a triple differences strategy, we provide evidence that both DUI arrests and alcohol related fatal traffic accidents fell, while alcohol related arrests increased, as a result of the expanded hours of Metro operation. However, we find that these effects may be due, in part, to individuals shifting their drinking to evenings when the Metro offered late night service from other evenings. Furthermore, we provide strong evidence that these effects were localized to areas close to Metro Stations and may reflect spatial shifting. Given evidence of both temporal and geographic shifting, the overall effects of public transportation provision on drinking and DUI behaviors on the entire DC area may be small.