The diets of 746 colon cancer cases in Los Angeles County, California (USA) were compared with those of 746 controls matched on age, sex, race, and neighborhood. In both genders, total energy intake was associated with significantly increased risk, and calcium intake was associated with significantly decreased risk. These effects were reduced only slightly after adjustment for the nondietary risk factors (weight, physical activity, family history, and, if female, pregnancy history). In men, total fat and alcohol intakes were responsible for the calorie effect; in women, no individual source of calories was associated independently with risk. Neither saturated fat nor fat from animal sources was responsible for the fat effect. There were no additional independent significant effects for sucrose, fiber, cruciferous vegetables, beta-carotene, other vitamins, or any other nutrient or micronutrient. In univariate analyses, meats, poultry, breads, and sweets were associated with excess risk, and yogurt was protective. After adjustment for sources of calories, no individual food was associated with excess risk, but yogurt remained significantly protective. Total calories were associated with excess risk throughout the colon while the effects of calcium, fat, and alcohol appeared somewhat stronger in the distal colon. After adjustment, crude fiber was significantly protective in the ascending colon but not even weakly protective in the distal colon.