To evaluate the accuracy of radiologic methods in liver metastases from colo-rectal carcinoma, a prospective investigation of 71 patients was performed. These patients were examined by angiography, ultrasound (US), computed tomography (CT) and computed tomographic angiography (CTA). The primary aim of the investigation was to determine the accuracy of CTA. The results of these examinations were then compared with the results from inspection and palpation of the liver at laparotomy. An analysis of liver tumor distribution, as well as a lesion-by-lesion analysis, was performed. In 20 patients, tumor growth was found in the liver in 36 lobes/segments at laparotomy. Three patients called false positives (angiography 2, US, CT and CTA 3 patients) turned out to be true positives since the lesions were overlooked at operation. The changes in sensitivity obtained when these patients are considered are given in parentheses. Of the lobes/segments affected by tumor growth at surgery, angiography revealed 47 (48) per cent, US 69 (71) per cent, CT 80 (82) per cent and CTA 83 (84) per cent. At a lesion-by-lesion analysis, angiography showed 33 per cent, US 69 per cent, CT 61 per cent and CTA 76 per cent of the lesions. Because of a larger number of false positives recorded with CTA compared with CT the former examination is not suitable as a single method for evaluating resectability of liver metastases.