Epidemiologic evidence on the relationship between nutrition and lung cancer is reviewed. Observational studies of diet and lung cancer, both prospective and retrospective, continue to suggest strongly that increased vegetable and fruit intake is associated with reduced risk in men and women; in various countries; in smokers, ex-smokers, and never-smokers; and for all histologic types of lung cancer. Prospective studies of blood beta-carotene levels, arguably the best available biomarker of vegetable and fruit intake, indicate that low levels are predictive of increased lung cancer incidence. However, in a randomized, placebo-controlled clinical trial in male smokers, lung cancer incidence and total mortality were increased significantly among the men receiving beta-carotene supplements. If beta-carotene can prevent lung carcinogenesis, which the trial cannot rule out, then the dosage, duration of use, method of administration, and/or subpopulation are critical. Ongoing clinical trials, some of which include women, will provide much-needed information. Other carotenoids, other phytochemicals, and associated dietary patterns may explain the beneficial effects of vegetables and fruits and have not been explored adequately in epidemiologic work. Several observational epidemiologic studies, both prospective and retrospective, have indicated that diets high in fat, saturated fat, and cholesterol may increase the risk of lung cancer and that the effect is not mediated through vegetable and fruit intake. The relationship, although not yet established, merits further investigation. Since beta-carotene can function as an antioxidant, other micronutrients with this potential, specifically vitamins E and C and selenium, also have been proposed to reduce lung cancer risk. However, the totality of the epidemiologic evidence is not, at present, persuasive for any one of these micronutrients.