We report here the results of the first multiinstitutional prospective evaluation of patients considered to have potentially resectable hepatic metastases from colorectal carcinoma. One hundred fifty-six patients were enrolled from 15 institutions. Six patients were subsequently excluded. One hundred fifty patients underwent surgery and are evaluable for analysis (median follow-up time, 3.1 years; range, 4 months to 5.1 years). Curative resection could be performed on 46% of patients (69 of 150), noncurative resection on 12% (18 of 150), while 42% were found to be unresectable (63 of 150). Thirty-day surgical mortality and morbidity rates in patients with attempted resection were 2.7% and 13%, respectively. The curative resection group was observed to have an improved median survival (37.1 months) compared with the noncurative resection group (21.2 months) and the unresectable group (16.5 months) (P less than .01). Computed tomographic (CT) scan was a poor predictor for resectability, and age was not a contraindication to curative resection. Preoperative carcinoembryonic antigen (CEA) values were also a poor predictor for resectability. However, the median CEA value 61 to 180 days postsurgery was significantly higher in unresectable patients compared with median CEA levels in noncuratively and curatively resected groups (P less than .01). Our results imply that curative resection leads to an increase in median survival. Noncurative resection provides no benefit to asymptomatic patients, since unresectable and noncurative resection groups have similar life expectancies. Longer follow-up will be needed to demonstrate the ultimate impact of curative resection on survival.