There is a well documented J-shaped relationship between alcohol consumption and mortality that is not fully understood. Differences between studies, particularly in the use of cigarette smoking as a covariate, may partially explain inconsistencies in findings. This study sought to clarify the nature of this alcohol-smoking-mortality relationship. The Lung Health Study enrolled 5,887 (3,702 male; 95% white) cigarette smokers with airways obstruction, aged 35 to 60. Measures of smoking, drinking and body mass index were available annually over the 5-year duration of the study, and pattern of alcohol use and diastolic blood pressure were assessed at baseline. Heavy drinkers were excluded from the sample. Documented evidence of hospitalizations was used in establishing morbidity. Proportional hazards regression was used for analysis. A significant protective effect of moderate drinking was found among men, but not women. There was also no apparent interaction between smoking status and drinks per week (both time dependent) in these data on morbidity and mortality. A baseline pattern of drinking eight or more drinks per occasion, but less than once a month, among men was associated with fewer hospitalizations and deaths. This study confirmed the general relationships between use of alcohol and illness or death, but with a few unexpected results. Further investigation is needed to identify whether these unexpected findings were due to our somewhat unique sample, to our use of hospitalization as the primary outcome measure, or to other causes.