One hundred eighty-nine surgically resected hepatocellular carcinomas (HCC) were analyzed to study tumor encapsulation and the pathologic features that might account for the better prognosis in relation to it, and to examine the prognostic and pathobiologic significance of capsular thickness. Tumor encapsulation was found in 72 (46.8%) of the 154 cases with adequate histologic sections of the tumor-nontumor junctions. Encapsulated tumors showed a much lower incidence of direct liver invasion (P less than 0.0001), tumor microsatellites (P less than 0.0001), and venous permeation (P = 0.02) when compared with nonencapsulated ones. Significantly better disease-free and actuarial survival times were observed in patients with encapsulated tumors (medians, 9.9 and 18.3 months, respectively), compared with those with nonencapsulated ones (medians, 4.0 and 5.9 months, respectively; P = 0.0001 and 0.001, respectively). The incidence of tumor encapsulation did not increase or decrease with tumor size. Tumor encapsulation did not correlate with the presence of cirrhosis or the abundance of tumor stroma, suggesting that formation of the tumor capsule was independent of the degree of fibrosis within and outside the tumor. Among the 72 cases of encapsulated HCC, the capsular thickness ranged from 0.13 to 3.09 mm (mean +/- standard deviation = 0.87 +/- 0.59 mm), and it was unrelated to tumor size or presence of cirrhosis. Although it was apparent that a lower extensive tumor invasiveness contributed significantly to the better prognosis in encapsulated HCC, there was no correlation between capsular thickness and liver invasion, microsatellites, venous permeation, or survivals. Therefore, the thickness of tumor capsules was not helpful in prognostication.