This study evaluates prospectively the reliability of computerized tomography (ct) in the pre‐operative staging of 80 patients with rectal cancer. The scans were performed and reviewed by one radiologist without knowledge of either the operative findings or the final clinicopathological stage of the tumours.Fourteen of 60 patients (23%) who had a potentially curative resection were correctly staged by ct. The tumour was understaged in 28 patients (47%) and was overstaged in 18 patients (30%). Ct did not identify the one patient with histological demonstration of invasion of an adjacent organ and was equivocal or incorrect in 10 others on the, question of adjacent organ invasion. Ct failed to define accurately local tumour spread confined to the rectal wall (positive predictive value (ppv) 23%). Identify venous invasion (ppv 35%) or involved regional lymph nodes (ppv 42%). However, the negative predictive value for excluding synchronous liver metastases was 90%, and 11 patients who subsequently developed histologically confirmed local recurrence were all correctly diagnosed on ct.These findings suggest that pre‐operative examination of patients with rectal cancer by ct is not routinely justified specifically for purposes of staging the dise