In a prospective study involving 833 consecutive outpatient and open-access colonoscopies, attempts were made to characterize the benefit of colonoscopy in terms of both predicted and unpredicted findings and therapeutic procedures. The endoscopist therefore predicted the endoscopic findings before the endoscopy. The results were compared for the different indications for colonoscopy. The overall agreement between the predictions and the colonoscopic findings was 61%. Clinically significant abnormalities were found in about half the examinations. The most frequent abnormal findings were benign polyps (24%), inflammatory bowel disease (17%), and malignancy (5%). In about half the patients with a malignancy the indication for colonoscopy was rectal bleeding, and half of the malignancies were not predicted. The greatest benefit of colonoscopy was found in patients referred because of overt rectal bleeding or occult faecal blood, and abnormal barium enema or endoscopy findings. The importance of complete colonoscopy in connection with operation for colorectal carcinoma is emphasized.