In a prospective cohort study of 1271 Massachusetts residents 66 years of age or older, we examined the association between consumption of carotene-containing vegetables and subsequent five year mortality. Dietary information was obtained by food frequency questionnaire in 1976. The relative risk of cancer mortality was examined within quintiles of green and yellow vegetable score (calculated from intake of carrots or squash, tomatoes, salads or leafy vegetables, dried fruits, fresh strawberries or fresh melon, and broccoli or brussel sprouts). After controlling for age and smoking behavior, those in the highest quintile of intake of these carotene-containing vegetables had a risk of cancer mortality which was 0.3 (95% confidence limits 0.10-0.96) that of those in the lowest quintile. The trend of decreased cancer risk with increasing intake of carotene containing vegetables was significant (p = .026). This relationship is consistent with the hypothesis that carotene may act as an inhibitor of carcinogenesis.