A total of 417 lung cancer cases and 849 controls were interviewed on their life-long tobacco usage and their current intake of four food items rich in retinol or carotene. The study was a hospital-based case control where 'cases' were lung cancer patients diagnosed during the period 1979/80 at seven hospitals in the Lombardy region (90% pathologically confirmed) and controls were patients admitted to the same hospitals for causes unrelated to tobacco smoking (epithelial cancers being excluded from present analysis). Odds ratios (OR) have been computed for increasing frequencies of consumption of liver, cheese, carrots and leafy green vegetables, having controlled for the confounding effects of tobacco usage, residence and birthplace. Current smokers who did not consume carrots showed a three-fold risk of developing lung cancer compared with those who ate them more than once a week (OR = 2.9 less than p less than 0.01); the ORs for consumers in the categories of 1-2 and 3-4 times per month were 1.8 and 2.0 respectively, with a significant test for linear trend (p less than 0.01). Among ex-smokers or non-smokers, no decrease of lung cancer risk is evident associated with carrot consumption. An excess risk was also associated with low intake of green vegetables although it was not significant, while no excess risk was evident for non-consumers of liver and cheese. The effect of carrots is independent of histological type of lung cancer while the effect of green vegetables was confined to epidermoid carcinomas: low versus high intake group OR = 1.3.