In his invited commentary about our article (1) in this issue of the Journal, Dr. Blot (2) emphasized the indisputable carcinogenicity of alcohol based on epidemiologic and laboratory evidence, regardless of the lack of classic genotoxicity data (1). He concluded that the data presented in our article (2) added to a large body of epidemiologic evidence which incriminates heavy alcohol use as a risk factor for upper aero-digestive tract cancers with or without concomitant exposure to tobacco smoking. Although we wholeheartedly share Dr. Blot's views on this issue, our argument was not nearly as cogent as his. We welcome the opportunity to comment on a couple of points he mentioned concerning our study and to discuss the potential effects of selection bias on the magnitude of the relative risk estimates for alcohol consumption in our case-control study (2). We agree with Dr. Blot that we have interpreted the interaction between smoking and alcohol somewhat liberally. We believe that both the direction and intensity of interaction may be a function not only of the anatomic subsite but also of the type of alcoholic beverage. The different levels of the cumulative alcohol consumption variable used in our study were qualitatively heterogeneous with respect to the relative contribution of the different types of beverage consumed by our subjects. In Brazil, the group with high cumulative consumption is more likely to be composed of drinkers of cachaga, a sugar cane distillate, and spirits, whereas low or intermediate cumulative consumption levels may have included more beer or wine drinkers. Assuming that the net carcinogenic effect of alcohol drinking may include both the exposure to alcohol per se and a component due to exposure to other substances present in the beverages, it is conceivable that the overall joint effect given concomitant smoking