Computerized tomography (CT) of liver is used in oncologic practice for staging tumors, evaluating response to treatment, and screening patients for hepatic resection. Because of the impact of CT liver scan on major treatment decisions, it is important to assess its accuracy. Patients undergoing liver transplantation or resection provide a unique opportunity to test the accuracy of hepatic-imaging techniques by comparison of findings of preoperative CT scan with those at gross pathologic examination of resected specimens. Forty-one patients who had partial hepatic resection (34 patients) or liver transplantation (eight patients) for malignant (30 patients) or benign (11 patients) tumors were evaluable. Eight (47%) of 17 patients with primary malignant liver tumors, four (31%) of 13 patients with metastatic liver tumors, and two (20%) of 10 patients with benign liver tumors had tumor nodules in resected specimens that were not apparent on preoperative CT studies. These nodules varied in size from 0.1 to 1.6 cm. While 11 of 14 of these nodules were <1.0 cm, three of 14 were >1.0 cm. These results suggest that conventional CT alone may be insufficient to accurately determine the presence or absence of liver metastases, extent of liver involvement, or response of hepatic metastases to treatment.