Aims  To study the course of male alcohol abuse from age 20 to age 70–80 years.

Design  A prospective multi‐disciplinary follow‐up of two community cohorts of adolescent males from 1940 until the present.

Setting and participants  Two hundred and sixty‐eight former Harvard undergraduates (college sample) and 456 non‐delinquent, socially disadvantaged Boston adolescents (core city sample).

Measurements  Since adolescence these cohorts have been followed by repeated interview, questionnaires and physical examination. The college cohort has been followed until age 80 and the younger core city cohort until age 70. DSM‐III criteria were used to ascertain alcohol abuse and alcohol dependence. At some point during their lives, 54 (20%) of the college men and 140 (31%) of the core city men met criteria for alcohol abuse. Outcome categories were mortality, continued alcohol abuse and stable remission.

Findings and conclusions  These socially divergent cohorts resembled each other in four respects. First, by age 70 chronic alcohol dependence was rare; this was due both to death and to stable abstinence. By age 70, 54% of the 72 successfully followed alcohol‐dependent core city men had died, 32% were abstinent, 1% were controlled drinkers and only 12% were known to be still abusing alcohol. By age 70, 58% of the 19 successfully followed college alcohol‐dependent men had died, 21% were abstinent, 10.5% were controlled drinkers and only 10.5% were known to be still abusing alcohol. Secondly, in both samples alcohol abuse could persist for decades without remission, death or progression to dependence. Thirdly, among both samples prior alcohol dependence and AA attendance were the two best predictors of sustained abstinence. Fourthly, few life‐time symptoms of alcohol abuse were the best predictor of sustained return to controlled‐drinking.