The effect of alcohol consumption was assessed in 561 men with myocardial infarction and 643 healthy controls recruited from France and Northern Ireland between 1988 and 1991 in the ECTIM Study (Enquête Cas-Témoins de I'Infarctus du Myocarde). In total, patients consumed less wine than did controls, while non-wine-derived alcohol consumption did not differ significantly. After adjustment for cardiovascular risk factors and country of recruitment by logistic regression, alcohol consumption displayed a protective effect against myocardial infarction, the magnitude of which was comparable in both countries. This effect, which was essentially due to wine consumption in France and to nonwine consumption in Northern Ireland, was largely attenuated by the introduction of high density lipoprotein cholesterol into the model. Thus, both wine and nonwine consumption appear to exert a protective effect against myocardial infarction which is partly mediated through an increase in high density lipoprotein cholesterol.