A cohort of 5,080 men living in a retirement community in California (United States) and initially free from lung cancer were followed from June 1981 to December 1989. At recruitment, each study participant completed a mailed questionnaire which requested information on the subject's medical history, use of cigarettes, and usual consumption frequencies during the preceding 12 months of 44 vegetable and fruit items. Men who had never smoked had the highest mean daily intake of f3-carotene (8,505 ~g), followed by past smokers (7,761 p.g) and then by current smokers (6,178 ~g). f3-Carotene intake of the subject's wife was correlated significantly with that of the husband in the 4,018 spouse pairs (r = 0.46; P = 0.0001). Among men with similar smoking habits, dietary 13-carotene intake significantly decreased with the spouse's smoking habit: never, past, and current smokers (P= 0.004; test for linear trend). During 31,477 person-years of follow-up, 125 incident cases of lung cancer were observed among the cohort of 5,080 men. Age-adjusted relative risks for lung cancer were below unity (i.e., demonstrating a reduced risk) for higher relative to lower consumption of 13-carotene, of all vegetables and fruits, and of yellow vegetables alone. However, these relative risks approached or crossed the null value when adjusted for personal smoking.