OBJECTIVE: To investigate by computerized morphometry the morphological features in hepatocellular carcinoma (HCC) expressing cytokeratin 7 (CK7)-positive and in HCC CK7-negative. STUDY DESIGN: This study included 15 HCC CK7- positive and 18 HCC CK7-negative liver transplant specimens from patients with hepatitis C virus (HCV)- related cirrhosis. All specimens were stained with hematoxylin-eosin and immunohistochemically for CD34 to assess the degree of sinusoidal capillarization, reticulin to assess the cell plate architecture, and Ki67 to identify neoplastic cells that are actively dividing. We generated a computerized morphometric model to evaluate the volume fractions occupied by hepatocyte nuclei and cytoplasm, sinusoids, portal triads, capillarized sinusoids, and tumor cells actively dividing. Lastly, the surface fraction occupied by reticulin was calculated. Moreover, tumor cells expressing both CK7 and Ki67 in CK7-positive HCC were also identified. Ten high-grade dysplastic nodules from resection specimens of HCVrelated cirrhotic livers were used as a control group. RESULTS: On hematoxylin-eosin stains the features most discriminatory between HCC CK7-positive and CK7-negative were volume fractions of sinusoids and of fibrosis and inflammatory infiltrate, which were significantly highest in CK7-positive HCCs. On immunohistochemistry, volume fractions of capillarized sinusoids and of Ki67 cells were significantly highest in CK7-positive HCCs. CONCLUSION: Our morphometric model is an objective method of quantification of the morphologic features in both CK7-positive and CK7-negative HCCs, and it could be applied in studies involving histological evaluation of the different subtypes of HCC arising in the cirrhotic liver according to CK7 expression. © Science Printers and Publishers, Inc.