Studies of isozyme composition in the rat liver-hepatocellular carcinoma model system have revealed wide-ranging abnormalities of gene expression. Isozymes geared for adult liver function are lost in tumors to varying degrees, depending on growth rate and degree of tumor dedifferentiation; whereas isozymes low or absent in normal adult liver become predominant or sole forms in fast growing, poorly differentiated hepatic tumors. The prevailing pattern is a switch from the adult to fetal forms, thereby indicating that genes coding for adult isozymes are suppressed in liver neoplasms, while genes active in fetal stages but inactivated during normal embryonic development become re-activated in cancer. These isozyme alterations not only are a key to neoplastic behavior, but are a striking example, among many, of pervasive abnormalities of gene expression involving virtually every means of identification of gene products; for example, the expression in tumors of fetal antigens, growth factors, angiogenesis factors, membrane components, and ectopic polypeptide hormones. Most important is the recently discovered activation of normally silent host oncogenes associated with and probably causally related to viral or chemical carcinogenesis. Despite compelling evidence for somatic mutation, these findings argue for abnormal gene regulation in cancer rather than gene mutation. Uriel has proposed that retrodifferentiation to an embryonal stage is a normal response to cell injury, and is an essential step preceding normal re-ordering of gene regulation. Under the influence of a carcinogen, however, normal differentiation may be aborted before differentiation is complete and the resultant transformed cell would then not be subject to normal regulatory restraints. A similar hypothesis has been expressed by Potter in the aphorism "Oncogeny is blocked ontogeny." Evidence exists for a role of 5-methylcytosine as a regulatory site in gene transcription. Jonathan Nyce, in a Ph.D. thesis presented to the Biology Department of Temple University, has proposed a mechanism of chemical carcinogenesis, based on the loss of cytosine residues, and therefore of methylation sites, in DNA as a result either of mispairing of bases resulting from O6-alkylation of guanines in DNA: or by inhibition of cytosine methylation. This plausible proposal provides a molecular basis for the many aberrations of gene regulation in cancer.