Some methodological issues concerning this study need attention. First, we used coronary calcification, a measure of coronary atherosclerosis,7 as the outcome. Because most individuals with coronary calcification are asymptomatic, changes in drinking habits elicited by clinical symptoms were not likely to affect the associations. Second, with increasing age, most drinkers reduce their level of alcohol consumption.17 In this population of asymptomatic older adults, less than a quarter of individuals consumed more than 2 alcoholic drinks per day. Current levels of alcohol consumption may not reflect the possibly higher level of consumption during earlier decades. If a moderate consumption of alcoholic beverages has a protective effect on atherosclerosis and if previously heavy drinkers report moderate alcohol use, the effect of moderate alcohol consumption may have appeared less protective. Similarly, if previously light to moderate drinkers report abstaining from alcohol, the protective effect of moderate alcohol consumption may be underestimated. Because of the limited range of alcohol consumption in the study population, we could not examine the relation between heavy drinking and coronary calcification. Third, nondrinkers may not be the most appropriate reference category, since this category consists of lifelong abstainers and former drinkers.18 Lifelong abstainers could have an adverse risk profile whereas former drinkers may have stopped drinking because of ill health, particularly ischemic heart disease. Thus, nondrinkers as the reference category has been suggested to exaggerate the apparent benefits of light to moderate alcohol consumption.18 However, bias in the category of nondrinkers probably did not play a considerable role in our study: of the 15.8% of individuals who did not drink alcohol at the time of scanning, 38.2% were former drinkers. Most lifelong nondrinkers (73.0%) were women, for whom abstaining from alcohol is common. Of the former drinkers, only 4% had been moderate or heavy drinkers in the past, and individuals with a history of coronary heart disease were excluded from our study. Fourth, self-reported drinking habits may result in underreporting of alcohol consumption, especially in heavy drinkers.19 Underreporting the level of consumption tendsto weaken the associations found. Bias by differential misreporting of alcohol consumption by sex is possible. However, because analyses were adjusted for sex, we do not expect this bias to fully explain our findings. Finally, drinking patterns may influence the association between alcohol intake and coronary heart disease.20- 21 Regularity in alcohol drinking was not ascertained.