The proportion of colorectal cancer attributed to dietary habits is high, but several inconsistencies remain, especially with respect to the influence of some food groups. To further elucidate the role of dietary habits, 1,225 subjects with cancer of the colon, 728 with cancer of the rectum and 4,154 controls, hospitalized with acute non-neoplastic diseases, were interviewed between 1992 and 1996 in 6 different Italian areas. The validated food-frequency questionnaire included 79 questions on food items and recipes, categorised into 16 food groups. After allowance for non-dietary confounding factors and total energy intake, significant trends of increasing risk of colorectal cancer with increasing intake emerged for bread and cereal dishes (odds ratio [OR] in highest vs. lowest quintile = 1.7), potatoes (OR = 1.2), cakes and desserts (OR = 1.1), and refined sugar (OR = 1.4). Intakes of fish (OR = 0.7), raw and cooked vegetables (OR = 0.6 for both) and fruit other than citrus fruit (OR = 0.7) showed a negative association with risk. Consumption of eggs and meat (white, red or processed meats) seemed uninfluential. Most findings were similar for colon and rectum, but some negative associations (i.e., coffee and tea, and fish) appeared stronger for colon cancer. Our findings lead us to reconsider the role of starchy foods and refined sugar in light of recent knowledge on the digestive physiology of carbohydrates and the insulin/colon cancer hypothesis. The beneficial role of most vegetables is confirmed, with more than 20% reduction in risk of colorectal cancer from the addition of one daily serving.