For a series of 568 married white men who died from coronary heart disease (CHD) and an equal number of matched controls, information was collected on a large number of variables, including daily alcohol consumption. The crude matched pair ratio estimate for any versus no daily drinking was 0.6 (95% confidence limits 0.4 to 0.7). After controlling for additional confounding variables the risk ratio for any versus no daily alcohol consumption was 0.6 (0.5--0.8). This preventive effect was restricted to light drinkers, defined as those who drank less than or equal to 59.2 ml (2 oz) of alcohol daily. These data provide strong evidence against a causal role of daily alcohol consumption in fatal CHD and are consistent with a small preventive effect of any versus no daily drinking which is attributable only to light but not heavy drinkers.