Food intake during the preceding 15 years was evaluated in detail in 41 patients treated for colorectal cancer and an equal number of matched control subjects by means of a dietary history technique that permitted quantitation of nutrients. Dietary habits of the control group could be compared against two larger groups of 371 hospital control and 430 population control subjects. Patients with cancer, who were interviewed after complete recovery from surgery, consumed more fat, protein, and carbohydrates, and thus more energy, than control subjects although these differences were not statistically significant. Per unit energy, the habitual diet of patients with cancer contained less cereal fiber (P less than 0.001), less riboflavin (P less than 0.05), less calcium (P less than 0.05), and less phosphorus (P less than 0.05) than the diet of the control subjects. A high intake of either cereal fiber, total fiber, calcium, and phosphorus in relation to energy intake was found to be associated with a reduced risk ratio of colorectal cancer. For colon cancer separately, a high intake of calcium and cereal fiber was associated with a reduced risk ratio. For rectal cancer, a high intake of total fiber and cereal fiber was associated with a reduced risk ratio. High alcohol consumption correlated with an increased risk ratio. These data are compatible with previous Scandinavian studies relating food consumption to the incidence and mortality of colorectal cancer.