High alcohol consumption is one of the major risk indicators for premature death in middle-aged men. An indicator of alcohol abuse--registration with the social authorities for alcoholic problems--was used to evaluate the role of alcohol in relation to general and cause-specific mortality in a general population sample. Altogether 1,116 men (11%) out of a total population of 10,004 men were registered for alcoholic problems. Total mortality during 11.8 years' follow-up was 10.4% among the non-registered men, compared to 20.5% among men with occasional convictions for drunkenness and 29.6% among heavy abusers. Fatal cancer as a whole was not independently associated with alcohol abuse, but oropharyngeal and oesophageal cancers together were seven times more common in the alcohol-registered groups. Total coronary heart disease (CHD) was significantly and independently associated with alcohol abuse, but nearly all the excess CHD mortality among the alcohol-registered men could be attributed to sudden coronary death. Cases with definite recent myocardial infarction were not more common in the alcoholic population. A combined effect of coronary arteriosclerosis and heart muscle damage secondary to alcohol abuse is suggested. Other causes of death strongly associated with registration for alcohol abuse include pulmonary embolism, pneumonia and peptic ulcer, as well as death from liver cirrhosis and alcoholism. Of the excess mortality among alcohol-registered subjects, 20.1% could be attributed to CHD, 18.1% to violent death, 13.6% to alcoholism without another diagnosis and 11.1% to liver cirrhosis.