OBJECTIVE A multiinstitutional study was performed to evaluate the efficacy of CT during arterial portography for determining the resectability of hepatic tumors. The impact of findings from CT during arterial portography on patients' treatment (i.e., surgical vs nonsurgical) was assessed. In patients considered to have resectable tumors, the accuracy of CT during arterial portography for predicting surgical findings was also evaluated. METHODS A retrospective study was done of 404 patients from three institutions who had CT during arterial portography during the period 1985-1991 as part of preoperative staging to determine the resectability of hepatic tumors. The tumors included metastases from colorectal carcinoma in 197 patients (49%); other hepatic metastases, mostly from adenocarcinoma of the stomach, pancreas, and biliary tree in 123 (30%); and primary hepatocellular carcinoma in 84 (21%). Imaging results were correlated with results of percutaneous biopsy of at least one hepatic lesion in patients whose tumors were considered unresectable. In patients whose tumors were considered resectable, results were correlated with preoperative percutaneous biopsy (obtained in almost all cases) and pathologic examination of a surgical specimen (all cases). Although each case was considered individually, four criteria were used for resectability: (1) accessibility of all lesions to lobar or wedge resection that would yield clear margins, (2) anticipation that residual liver tissue after resection would provide sufficient function, (3) the absence of invasion of central hepatic vascular or biliary structures, and (4) the absence of extrahepatic disease. No specific restriction was made with respect to the number of hepatic lesions present. The accuracy of findings by CT during arterial portography for predicting resectability was assessed in the 146 patients who had tumors that were considered resectable on the basis of imaging findings and had surgery. RESULTS Of 404 patients, only 146 (36%) were thought to be candidates for resection on the basis of findings from CT during arterial portography. Of these, 122 (84%) actually had resection. The 24 patients who did not have resection included 22 patients with disease understaged or overstaged by CT during arterial portography, one with true-negative findings by CT during arterial portography, and one who died during surgery. The accuracy of findings by CT during arterial portography for predicting results at surgery was 85% for all patients and 91% for the subset of patients who had primary colorectal tumors with hepatic metastases. CONCLUSIONS Our experience shows that CT during arterial portography is a useful procedure for assessing the resectability of hepatic tumors. In our study, 64% of patients were spared unnecessary surgery.(ABSTRACT TRUNCATED AT 400 WORDS)