This study examined inter-beverage differences in the relationship between alcohol consumption and health status as reported by 17 249 respondents to the Canada Health Survey. Self-reported morbidity rates (bed-days, activity loss, doctor visits) were computed for consumers of beer, wine, liquor and for those with no specific beverage preference. These rates were compared with the level of expected morbidity based on each group's demographic attributes. Subsequently, dose-response relationships were described relating frequency and quantity of consumption to a standardized morbidity rate for each beverage-preference group. In general, the results supported the importance of inter-beverage differences as an intervening variable in the relationship of consumption to morbidity. Overall morbidity rates and both frequency and quantity dimensions of the dose-response relationship varied markedly as a function of type of beverage consumed. Beer drinkers, in particular, varied from other consumers they had significantly lower rates of morbidity than expected. Increases in frequency of beer drinking were associated with reductions in morbidity, but mildly deleterious effects were associated with excessive consumption.