Background: Neuroendocrine tumors commonly metastasize to the liver. Although surgical resection is considered a treatment option for patients with localized metastases confined to the liver, the longterm survival benefit of liver resection has not been dearly demonstrated. We examined the survival of patients undergoing liver resection for this disease. Study Design: Between 1984 and 1995, we evaluated 38 patients with liver-only metastases from neuroendocrine tumors, including 21 carcinoid, 13 islet cell, and 4 atypical neuroendocrine neoplasms. Data from a combined prospective and retrospective database and a tumor registry were analyzed. Of these patients, 15 underwent complete resection of all known disease. The remaining 23 patients, who also had disease confined to the liver, had comparable tumor burden but were believed to be unresectable. The longterm survival rates of these two groups were compared. Results: Patients who underwent liver resection did not differ from those who were unresectable with regard to age, pathology, primary tumor site, serum alkaline phosphatase levels, or percentage of the liver involved. All resections were complete, leaving no residual disease, and consisted of lobectomy (n = 3), segmentectomy (n = 1), and wedge resections (n = 11). There were no operative deaths. Patients who underwent hepatic resection had a significantly longer survival than unresected patients. Although median survival had not been reached in resected patients, the median survival in the unresectable group was 27 months. Patients who underwent liver resection had a higher 5-year actuarial survival (73% versus 29%). Conclusions: Hepatic resection in selected patients with isolated liver metastases from neuroendocrine tumors may prolong survival. This conclusion was reached by comparing our resected group with an unresectable group with similar tumor burden.