The relation of alcohol intake to total mortality is J-shaped. Abstainers have modestly higher mortality than moderate drinkers but considerably lower [corrected] mortality than heavy drinkers. The higher mortality among abstainers cannot be explained by selection or the presence of other risk factors. Known biologic mechanisms support the conclusion that moderate drinking increases the lifespan. No major differences have been found between the effects of beer, wine or liquor. While drinking patterns and changes in these influence mortality over time, little is known about their significance. The lowest risk of death seems to be at the average intake level of one drink per day. However, due to several sources of error in the assessment of alcohol intake no precise limits of optimal or safe drinking can be recommended. Trials are needed to ascertain these limits. Drinkers should practice moderation and watch for any harmful effects of alcohol.