Hepatoblastoma, a childhood tumor of the liver, is composed of epithelial and mesenchymal elements in varying proportions and at various stages of differentiation. The epithelial element recapitulates the stages of hepatocyte development from the primitive blastema through embryonal hepatocytes to fetal hepatocytes. The blastemal or undifferentiated cells have been postulated to represent neoplastic hepatocyte progenitor cells. In this study, we examine the immunophenotype of the various epithelial cells of hepatoblastoma with special emphasis on the small undifferentiated cell component and compare it with that of adult hepatocytes and hepatic stem (oval) cells. Putative stem cells in the liver can express all of the following markers: alpha-feto protein, CK19 (OV-6), chromogranin A, Bcl-2, HepPar-1, and alpha1 microglobulin. The latter, like alpha-feto protein, is a plasma protein synthesized by hepatocytes. Both alpha1 microglobulin and HepPar-1 are expressed in fetal liver cells as early as 7 weeks of intrauterine life. They are also expressed in hepatocellular carcinoma and in hepatocytic cell lines derived from normal fetal or adult liver. Formalin-fixed, paraffin-embedded archival tissues from 10 predominantly epithelial hepatoblastomas were immunostained with antibodies directed against CD 34, alpha1 microglobulin, Bcl-2, HepPar 1, and CK19 using the avidin-biotin-peroxidase method. The undifferentiated small cell component did not express any of the markers studied, namely, Bcl-2, HepPar-1, alpha(1) microglobulin, CD34, or CK19. Hepatocyte-like cells were alpha1 microglobulin- and HepPar-1-positive, with the intensity of staining correlating with the degree of hepatocytic differentiation. Bcl-2 expression was restricted to areas of ductular differentiation. CK19 was detected in foci that showed duct formation. The small cells of hepatoblastoma did not express HepPar-1, Bcl-2, CK19, alpha1 microglobulin, or CD34, markers that characterize the immunophenotype of hepatic stem cells ("oval" cells). Thus, this observation raises the following questions: (1) is "hepatoblastoma" a misnomer? (2) is the expression of tumor antigens dysregulated in hepatoblastoma? (3) does the liver have two different types of progenitor cells, oval cells and blastemal cells, with differing immunophenotypes? and (4) do the blastemal cells, rather than oval cells, represent the more primitive progenitor cells of the liver?