Using immunohistochemical demonstration of glutathione S-transferase placental type (GST-P) and histochemical demonstration of gamma-glutamyltransferase (gamma-GT), the long-term development of preneoplastic and neoplastic lesions was followed in rats over a 50-week period. Rats were given a single i.p. injection of 200 mg/kg body weight of diethylnitrosamine (DEN), and then 2 weeks later were administered 0.02% 2-acetylaminofluorene (2-AAF) (group 1), 0.05% phenobarbital (PB) (group 2), 2.0% butylated hydroxyanisole (BHA) (group 3) or no supplement (group 4) in their diet for 6 weeks, all rats being subjected to partial hepatectomy at week 3. Hepatocellular proliferated lesions were classified as foci, nodules and hepatocellular carcinomas. Development of foci, nodules and hepatocellular carcinomas was enhanced strongly by 2-AAF and weakly by PB, and inhibited by BHA. Almost all foci and nodules were GST-P positive, although 5-10% of the GST-P-positive foci were gamma-GT negative. The areas of GST-P-positive foci and nodules increased with time in all groups. In contrast, while the areas of gamma-GT-positive lesions also increased with time in groups 2-4, they decreased from week 12 in group 1. As the percentage gamma-GT-positive area in GST-P-positive foci significantly decreased with time in all groups, the rate of phenotypic reversion of gamma-GT in foci in group 1 was revealed to be larger than the focus growing rate, whereas that in groups 2-4 was smaller. Gamma-GT-negative and GST-P-positive micro-nodules of altered morphology appeared within gamma-GT- and GST-P-positive nodules in later stages. All hepatocellular carcinomas found in this experiment consisted of GST-P-positive cells. In contrast, 37% (13/35) of the hepatocellular carcinomas were negative for gamma-GT. The results indicate GST-P to be the most accurate marker enzyme for detection of initiated cells during liver carcinogenesis and gamma-GT to be more appropriate for indicating changes of phenotypic expression in each lesion type.