Findings in more than 30 population studies have suggested the relationship between alcohol consumption and mortality rates to be described by a U-shaped or J-shaped curve. The downstroke of the U is probably to be explained by a beneficial effect of moderate alcohol consumption on the risk of ischaemic heart disease. Various sources of bias in our interpretation of the downstroke of the curve have been reviewed and to some extent excluded. In all likelihood, however, both the drinking pattern and the type of alcohol used are determinants of the shape of the curve. One study has shown steady drinking to be more protective than binge drinking against heart disease, and a couple of studies have suggested wine-drinkers to enjoy greater protection than beer and spirits drinkers. Moreover, it is emphasised that the effect of alcohol on the risk of cardiovascular and other diseases is highly dependent on the distribution of other risk factors in the population under study. The pultative beneficial effects of moderate alcohol consumption should therefore be considered to apply only to individuals at risk of cardiovascular disease. Finally, it is stressed that available evidence of the efficacy of alcohol as a "heart medicine" is insufficient.