We have reviewed 156 papers which provided sufficient information to relate individual alcohol consumption to risk for a variety of physical damage. Overall, there was evidence for a dose-response relationship between level of alcohol consumption and risk of harm for liver cirrhosis, cancers of the oropharynx, larynx, oesophagus, rectum (beer only), liver and breast, and blood pressure and stroke. An increased risk of cardiac arrhythmias, cardiomyopathy and sudden coronary death was associated with heavy drinking. There was evidence for a protective effect of alcohol consumption against risk of coronary heart disease, which could be achieved at consumption levels of less than 10 g alcohol a day. The mortality of non-drinkers was higher than that of moderate drinkers in some studies. Level of alcohol consumption and total mortality were dose-related when non-drinkers were excluded. The finding of a dose-relationship between alcohol and harm suggested causality. It was not possible to define individual risk for all harms at a given level of alcohol consumption because of variations in methodology, but some idea of the order of magnitude of the increased risk can be obtained from calculating trends of pooled log-odds ratios. At levels of alcohol consumption of more than 20-30 g a day, all individuals are likely to accumulate risk of harm. Current guidelines on upper limits of lower risk drinking in different countries (168-280 g of alcohol a week for men and 84-140 g a week for women) reflect levels at which the risk of total mortality is not greatly increased above one.