BACKGROUND: Early-onset drunkenness is associated with an increased risk of developing an alcohol use disorder (AUD), which predicts excess mortality risk. Here, we estimated mortality risk for drinkers with and without early drunkenness. METHODS: For 14,848 adult participants interviewed about drinking, drunken episodes, and AUD in 1981-83 for the Epidemiologic Catchment Area in New Haven (Connecticut), Baltimore (Maryland), St. Louis (Missouri), and Durham (North Carolina), we linked National Death Index records through 2007. RESULTS: Cox regression modeling estimates showed excess mortality for drinkers with age of first drunkenness earlier than 15 years old (hazard ratio, HR: 1.47, 95% CI: 1.25, 1.72) and when first drunkenness occurred at or after age 15 (HR: 1.20, 95% CI: 1.11, 1.29), as compared with adults who had never been drunk. Consistent results were observed, irrespective of AUD history. That is, early drunkenness signaled excess mortality risk even in absence of AUD. CONCLUSIONS: In a large community sample from four cities in the US, early age of onset of drunkenness predicts mortality risk. We discuss experiments to investigate the possible causal significance of this predictive association.