Computed tomography with rectal air insufflation was compared with transrectal ultrasonography (TRUS) in 63 patients. The CT protocol involved pre- and postcontrast scans with 5 mm slice thickness following air insufflation in IV antiperistaltic agent. Of the patients, 79 % were scanned in the prone position. Results of the preoperative examinations were compared with the histological findings. The CT examination had an accuracy rate of 74 %, predicting perirectal spread with a sensitivity of 83 % and a specificity of 62 %, whereas the corresponding figures for TRUS were 83, 91 and 67 %. The accuracy, sensitivity and specificity of CT and TRUS for nodal involvement were 57, 56, 57, 66, 68 and 64 %-respectively. These findings confirm that TRUS is more accurate than CT in local tumour (T) staging and in detecting nodal (N) spread. However, the appropriate CT technique shows spread of tumour outside the rectal wall and locoregional lymph nodes with reasonable accuracy. Lymphatic spread correlated with nodal size. TRUS and CT correctly staged only 57 and 43 %, respectively, of cases with nodal metastases with maximum diameter of 5 mm. TRUS sometimes overstaged perirectal growth of tumour in 7 patients, due to inflammation (5 patients) or incorrect positioning of the balloon in relation to the tumour surface (2 patients).