Conventionally, cohort studies have assessed the association between alcohol and all-cause mortality by using alcohol intake at enrolment.In the Melbourne Collaborative Cohort Study, participants were asked about usual frequency and quantity of beverage-specific alcohol intake for 10-year periods starting at age 20 from which current, past and lifetime intakes were calculated. We used Cox regression to estimate hazard ratios for mortality for 39 577 participants of the Melbourne Collaborative Cohort Study aged 40-69 at baseline.After a mean follow-up of 15 years/person, we identified 4639 deaths. Associations between all-cause mortality and lifetime, current (baseline) and past intake were J shaped, with lower mortality at low intake (e.g. <40 g/day for men and 10 g/day for women using lifetime intake) and elevated mortality at higher intake. For men, consistent light-to-moderate drinking (>0-39/>0-39 g/day) from age 20 to baseline age was associated with a 16% lower mortality, while heavy drinking at both ages (≥80/≥40 and ≥40/0 g/day) was associated with higher mortality compared with stable abstinence.Our findings support a reduced mortality risk associated with low-dose drinking but also highlight a higher mortality risk for consistent heavy drinking from a young age.