Results of many studies, summarized in this review, support the hypothesis that carcinogen-induced hepatocellular carcinomas develop from phenotypically-altered hyperplastic hepatocellular nodules; these in turn apparently arise from smaller focal collections of hyperplastic cells referred to as hepatocellular islands. The very recent recognition that phenobarbital, when administered after carcinogens, fosters the outgrowth of hepatocellular islands and carcinomas, now provides the means for studying stages of initiation and promotion in hep-atocarcinogenesis. In addition, the recognition that enzymatic alterations, particularly the acquisition of canalicular gamma glutamyl transpeptidase activity, loss of ATP'ase activity, and loss of glucose-6-phosphatase activity that characterize many islands, have been particularly useful for measuring and evaluating the growth kinetics and heterogeneity of the islands. Evidence is presented that periportal gamma glutamyl transpeptidase positive hepatocytes are considerably more abundant after four weeks of feeding .02% 2-acetyl-aminofluorene to young rats than in control animals, and that the outgrowth of these cells is fostered by a distinctive type of periportal reparative hyperplasia. The cells appear to arise from a pool of cells that are normally abundant in periportal location in young growing rats. The studies suggest that it may now be possible to develop short term in vivo bioassays for initiators, promoters, and complete carcinogens in the rodent liver.