A case-control study of cancer of the colon and rectum was conducted in the Marseilles region of southern France. Cases (399) and a corresponding number of controls, matched for age and sex, were included, the controls being selected from patients undergoing functional re-education for injuries or trauma which reduced their mobility. A dietary history questionnaire was used to determine the usual eating habits during the year preceding diagnosis for cases, or preceding interview for controls. The cases reported lower consumption of vegetables and oil than controls, but no differences were seen in the consumption of meat, bread, eggs or butter. The intake of several nutrients, particularly vitamins B2, B6, C, potassium, iron, magnesium and vegetable fibre, was lower among cases. However, when all these nutrients were analysed jointly and adjusted one for the other, only potassium retained a significant effect. This may be due to the high degree of colinearity between the estimated intake of many nutrients. No association was seen with fat or fibres from cereals.