In an effort to determine the reliability of colonoscopy the authors retrospectively reviewed preoperative colonoscopic findings and compared them with the postoperative pathologic specimen reports. Only lesions greater than 0.5 cm were included in the comparison. Over a 13-year period, 429 patients with colorectal cancer underwent preoperative colonoscopy. Four hundred thirteen (97 percent) of the colonoscopic examinations correlated with the pathologic specimen, but, in 16 cases (3 percent), lesions were missed. In total, 17 adenomatous polyps and 3 cancers were found in the surgical specimens that were not documented at colonoscopy. Eighteen patients had total preoperative colonoscopy and total abdominal colectomy, which makes for a reliable model to judge the accuracy of colonoscopy. In these 18 patients, 17 of the pathologic specimens correlated with the endoscopic findings, which yields an accuracy rate of 94 percent. Blind areas in the colon, plus misjudgment that the scope had reached the cecum, are responsible for the majority of colonoscopic errors.