Flavonoids are polyphenolic antioxidants naturally present in vegetable foods. Some flavonoids, such as quercetin, inhibit carcinogenesis in rodents, but their effect in humans is unknown. We measured the flavonoids quercetin, kaempferol, myricetin, apigenin, and luteolin in foods and assessed flavonoid intake in 1985 by dietary history in 738 men aged 65-84 years without a history of cancer, who were then followed for five years. Mean flavonoid intake was 25.9 mg/day. The major sources of flavonoid intake were tea at 61% and vegetables and fruits (mainly onions, kale, endive, and apples) at 38%. Between 1985 and 1990, 75 men developed cancer (all sites) and 34 men died from cancer. Flavonoid intake in 1985 was not associated with incidence of all-cause cancer (p for trend = 0.54) or with mortality from all-cause cancer (p for trend = 0.51). Flavonoid intake was also not associated with risk of cancers of the alimentary and respiratory tract (p for trend = 0.92). Adjustment for age, body mass index, smoking, physical activity, and vitamin C, vitamin E, beta-carotene, and dietary fiber intake did not change the relative risks. A high intake of flavonoids from vegetables and fruits only was inversely associated with risk of cancer of the alimentary and respiratory tract (relative risk of highest vs. lowest tertile = 0.51, 95% confidence interval 0.25-1.05); these results suggest the presence of other nonvitamin components with anticarcinogenic potential in these foods. We conclude that intake of flavonoids, mainly from tea, apples, and onions, does not predict a reduced risk of all-cause cancer or of cancer of the alimentary and respiratory tract in elderly men. The effect of flavonoids on risk of cancer at specific sites needs further investigation in prospective cohort studies.