Many epidemiological studies have focused on the relationships between diet and colorectal cancer, but only a few have been conducted in the Mediterranean area. A population-based case-control study was carried out from July 1987 to June 1989 in a low-risk area in Southern Italy. By means of an "ad hoc" tumor registry, 132 diagnosed colorectal cancers were detected during the two years of study. One hundred nineteen of these 132 colorectal cancer cases were interviewed about their personal dietary habits with use of a questionnaire concerning the frequency of consumption of 70 foods or beverages. An equal number of controls was randomly selected from the lists of general practitioners of the area during the same period and interviewed with the same food frequency questionnaire. In a multivariate analysis, the relative risks (RRs) of developing colorectal cancer were estimated according to the different levels of consumption of food groups and selected food items. All RRs were adjusted for age, sex, education, smoking status, and modifications in diet in the previous 10 years. The risk of colorectal cancer increased nearly threefold for the highest level of consumption of foods with a high content of refined sugar [RR = 2.75, 95% confidence interval (CI) 1.26-5.97] and for the consumption of wine (> 1 l/day) (RR = 3.22, 95% CI 1.05-9.88). An inverse relationship was revealed for the highest consumption of raw and cooked vegetables (RR = 0.51, 95% CI 0.25-1.04) and diary products (RR = 0.46, 95% CI 0.22-0.98) and for the consumption of more than two cups of coffee per day (RR = 0.38, 95% CI 0.16-0.89). In this Mediterranean area, the main source of calories, cereals, did not show a significant relationship with colorectal cancer. These findings support the hypothesis that the local Mediterranean dietary pattern could explain the low risk of colorectal cancer.