Many epidemiological studies have described a U-shaped relation between alcohol intake and all-cause mortality (Boffetta and Garfinkel, 1990; Fuchs et al., 1995; Gronbaek et al., 1994; Marmot et al., 1981). Most researchers attribute the 'U' to a combination of beneficial and harmful effects of ethanol itself. It has, on the other hand, been explained as an artefact due to misclassification or confounding (Shaper et al., 1998). Most of the studies of the effect of total alcohol intake have found that the descending leg of the curve mainly is attributable to death from cardiovascular disease (Rimm et al., 1991; Stampfer et al., 1988). Until recently, most studies addressed the effect of the three beverages taken together as ethanol. Studies of the correlation between wine intake per capita in different countries and incidence of ischaemic heart disease gave rise to the hypothesis that there is a a more beneficial effect of wine than of beer and spirits. Leger et al., Renaud and de Lorgeril and later Criqui and Rigel found an inverse relation between incidence rates of ischemic heart disease and wine consumption in different countries, but no such relation for the other types of beverages (Criqui and Rigel, 1994; Leger et al., 1979; Renaud and de Logeril, 1992).