To review prospective epidemiologic studies and randomized trials regarding the role of antioxidant vitamins (vitamins E and C and beta-carotene) in the prevention of cardiovascular disease, with emphasis on differences in results obtained by these two types of studies. Computerized and manual searches of the literature on antioxidant vitamins and cardiovascular disease. Prospective epidemiologic studies and randomized trials that included 100 or more participants and provided quantified estimates of antioxidant vitamin intake. Comparisons of relative risk reductions (RRR) across observational studies and randomized trials, including assessment of dose-response relations. All three large epidemiologic cohort studies of vitamin E noted that high-level vitamin E intake or supplementation was associated with a significant reduction in cardiovascular disease (RRR range, 31% to 65%), as measured by various fatal and nonfatal cardiovascular end points. To obtain these reductions, vitamin E supplementation must last at least 2 years. Less consistent reductions were seen in studies of beta-carotene (RRR range, -2% to 46%) and vitamin C (RRR range, -25% to 51%). Considerable biases in observational studies, such as different health behaviors of persons using antioxidants, may account for the observed benefit. By contrast, none of the completed randomized trials showed any clear reduction in cardiovascular disease with vitamin E, vitamin C, or beta-carotene supplementation. The trials were not specifically designed to assess cardiovascular disease, did not provide data on nonfatal cardiovascular end points, may have had insufficient treatment durations, and used suboptimal vitamin E doses. The completed trials were of adequate size to indicate that the true therapeutic benefit of vitamin E and other antioxidants in reducing fatal cardiovascular disease (a survival benefit as long as 5 years) is probably more modest than the epidemiologic data suggest. The epidemiologic data suggest that antioxidant vitamins reduce cardiovascular disease, with the clearest effect for vitamin E; however, completed randomized trials do not support this finding. Much of this controversy should be resolved by the ongoing large-scale and long-term randomized trials designed specifically to evaluate effects on cardiovascular disease.