Male Sprague-Dawley rats received the hepatocarcinogen N-nitrosomorpholine (NNM) in the drinking water at low dose levels ranging from 6 mg/l to 60 mg/l for 6 and 12 weeks, respectively. Foci of altered hepatocytes (FAH) were demonstrated histochemically using changes in the activities of glucose-6-phosphate dehydrogenase and glycogen phosphorylase, and in the glycogen content as markers. Proliferating cells were detected by the immunohistochemical reaction for proliferating cell nuclear antigen (PCNA). The number and size of foci of altered hepatocytes increased in a time and dose-related manner. The dose-effect curves were non-linear with a slight positive slope at the low doses and a markedly increased slope at higher doses. The number of PCNA positive hepatocytes showed a dose-dependent increase. In addition to the granular distribution of PCNA in the nuclei, hepatocyte nuclei with homogeneously distributed PCNA occurred in animals exposed to 60 mg/l NNM. It is proposed that these cells are related to the occurrence of hepatocytes with higher ploidy induced by NNM and may be regarded as cells in the G2 phase of the cell cycle. The non-linear shape of the dose-response-curve of the FAH suggests that some mechanisms contribute to carcinogenesis over the whole dose range, whereas other mechanisms enhance carcinogenesis only at higher doses. The relevance of the non-linear dose-effect curve for the risk assessment of carcinogens is discussed.