We conducted a population-based study of diet and lung cancer among the multiethnic population of Hawaii in 1983-1985. We completed interviews for 230 men and 102 women with lung cancer and 597 men and 268 women controls, frequency-matched to the patients by age and sex. A quantitative dietary history assessed the usual intake of foods rich in vitamins A and C and carotenoids. A clear dose-dependent negative association was demonstrated between dietary beta-carotene and lung cancer risk in both sexes. After adjusting for smoking and other covariates, the men in the lowest quartile of beta-carotene intake had an odds ratio of 1.9 (95% confidence interval, 1.1-3.2) compared to those in the highest quartile of intake. The corresponding odds ratio for women was 2.7 (95% confidence interval, 1.2-6.1). No clear association was found for retinol, vitamin C, folic acid, iron, dietary fiber, or fruits. All vegetables, dark green vegetables, cruciferous vegetables, and tomatoes showed stronger inverse associations with risk than beta-carotene. This observation suggests that other constituents of vegetables, such as lutein, lycopene, and indoles, and others, may also protect against lung cancer in humans.