Using newly available food composition data for carotenoids, the authors reanalyzed a population-based case-control study of diet and lung cancer conducted in Hawaii in 1983-1985 (L. Le Marchand et al., J. Natl. Cancer Inst., 81: 1158-1164, 1989). The analysis included interviews with 230 men and 102 women with lung cancer and 597 men and 268 women as controls, frequency-matched to the patients by age and sex. A previously validated quantitative diet history assessed the usual intake of foods rich in carotenoids. After adjusting for smoking and other covariates, no association was found with lung cancer risk for dietary lycopene or beta-cryptoxanthin intake, whereas dose-dependent inverse associations of comparable magnitude were found for dietary beta-carotene, alpha-carotene, and lutein. When subjects were cross-classified by their joint intakes of the latter three carotenoids, those who had a high intake (> median) for all three had the lowest risk for lung cancer. In a similar two-way interaction analysis, the previously reported inverse association of lung cancer with vegetable consumption in these data was found to be stronger than that with intake of these three carotenoids. Consistent with our previous findings, this analysis provides further evidence for a protective effect of certain carotenoids against lung cancer and for the greater protection afforded by consuming a variety of vegetables compared to only foods rich in a particular carotenoid.