The role of alcohol abuse in mortality was studied in an unselected population of over 10,000 46-48-year-old men in Malmö, Sweden. During follow-up of 0-6 years (mean 3 years) 199 men died. In 61 men (30.7%) death was alcohol related. A theoretical calculation of excess deaths in men with an alcohol-positive history yielded 78 deaths (39.2%). In the official cause of death statistics 10 of the deaths had been assigned alcoholic aetiology (5.0%). These estimates indicate that alcohol was the commonest underlying factor in death in this sample of middle-aged men. The number of deaths with alcoholic aetiology in official cause of death statistics should be multiplied by a factor of six to eight to arrive at the true alcohol-related death rate.