As an indicator of hospitalization costs, the number of hospitalized days among 82,430 white and black adult health maintenance organization (HMO) members was studied. Baseline data were collected at health examinations from 1978 to 1982. Using all hospitalizations from 1978 to 1983 that were not related to pregnancy and that occurred after examination, multiple regression analyses for each race-sex subgroup were controlled for alcohol use (eight categories), cigarette smoking (four categories), age (three categories), and period of observation. Estimates of hospital days associated with alcohol, age, and smoking were calculated for specific diagnostic categories and for all diagnoses by projecting observed days hospitalized of lifelong alcohol abstainers, persons less than 45 years old, and lifelong nonsmokers. Among men of both races, there was almost no increase in total hospital days associated with alcohol use when all levels of drinking were considered. Among women, because of the preponderance of lighter drinkers who had fewer hospitalizations, the overall percentage of total hospital days associated with alcohol was negative. Several trends were identified between alcohol use and hospital days for the diagnostic groups. Most notable were a deficit of hospital days for circulatory system problems and an excess of hospital days for all neoplasms. In summary, in this population of prepaid health-plan members, use of alcoholic beverages was only a minor factor in total hospitalization experience.