The mortality rate from hepatic failure after extensive resection should be negligible in the presence of normal results from preoperative liver function tests in patients without pre-existing hepatitis and cirrhosis. Despite conventionally acceptable results from preoperative hepatic function tests in 56 patients undergoing extensive hepatic resection for tumours (47 metastatic, six hepatomas and three adenomas), however, five patients died of hepatic failure. Among the many preoperative and intraoperative risk factors studied, the important factors in the group with hepatic failure were very high levels of serum alkaline phosphatase (p less than 0.05) in the presence of normal levels of bilirubin and large tumor, preoperative administration of chemotherapy, the presence of hepatomas rather than metastatic carcinoma (p = 0.083) and intraoperative blood loss of greater than 5,000 milliliters (p = 0.03). The patients receiving preoperative chemotherapy or those with hepatoma showed a minimal rise of alkaline phosphatase (p less than 0.03) and a minimal regeneration of liver on computed tomographic (CT) scan after hepatic resection. In the group with hepatic failure, a consistent postoperative pattern of increasing bilirubin with normal or subnormal alkaline phosphatase levels corresponded with lack of regeneration of liver on repeated CT scans. Conversely, the pattern of decreasing bilirubin with reciprocal increase in alkaline phosphatase corresponded with hepatic regeneration on CT scan in the group of survivors. Thus, we observe that alkaline phosphatase is a good indicator of hepatic regeneration in the absence of jaundice in patients after hepatectomy. To avoid postoperative hepatic failure, we recommend more discriminant tests than conventional hepatic function tests in patients with large tumors associated with high alkaline phosphatase levels, preoperative chemotherapy and hepatoma even without pre-existing cirrhosis or hepatitis.