To determine the relation between multivitamin use and death from heart disease, cerebrovascular disease, and cancer, the authors examined a prospective cohort of 1,063,023 adult Americans in 1982-1989 and compared the mortality of users of multivitamins alone; vitamin A, C, or E alone; and multivitamin and vitamin A, C, or E in combination with that of vitamin nonusers by using multivariate Cox proportional hazard models. Multivitamin users had heart disease and cerebrovascular disease mortality risks similar to those of nonusers, whereas combination users had mortality risks that were 15% lower than those of nonusers. Multivitamin and combination use had minimal effect on cancer mortality overall, although mortality from all cancers combined was increased among male current smokers who used multivitamins alone (relative risk (RR) = 1.13, 95% confidence interval (CI): 1.05, 1.23) or in combination with vitamin A, C, or E (RR = 1.16, 95% CI: 1.06, 1.26), but decreased in male combination users who had never (RR = 0.86, 95% CI: 0.74, 0.99) or had formerly (RR = 0.90, 95% CI: 0.82, 0.98) smoked. No such associations were seen in women. These observational data provide limited support for the hypothesis that multivitamin use in combination with vitamin A, C, or E may reduce heart disease and cardiovascular disease mortality, but add to concerns raised by randomized studies that some vitamin supplements may adversely affect male smokers.