Epidemiological studies have suggested an association between diet and colorectal cancer. Case/control studies, however, have been scarce, and studies based on interview with cancer patients who have symptoms from their cancer are inevitably prone to bias. An endoscopic population screening study for detection of colorectal adenomas enabled a double-blind registration of diet during 5 consecutive weekdays. Neither the participant nor the dietitian was informed of the findings at endoscopy. The estimation of 23 nutritional components was based on analysis of local commercial food and on the composition of foods in Norway. Results showed increasing consumption of fat and decreasing consumption of fiber and cruciferous vegetables in the presence of increasing neoplastic changes. The present material will form the basis for dietary-related follow-up studies.