We have been unable to demonstrate the unequivocal presence of transferrin receptors on rat hepatocytes. The binding and presumed internalisation of lZ5I-Fe2 +-transferrin by freshly isolated hepatocytes was only partially inhibited by up to a 104-fold molar excess of unlabelled ligand and was virtually insensitive to chloroquine. There would appear to be only a weak association between this ligand and some component of the hepatocyte cell surface. These results were not compatible with the commitment of Fe;+-transferrin to either a receptormediated endocytotic pathway nor to a rapid recycling pathway through sorting endosomes. Desialylation of the biantennary oligosaccharide side chains of Fe:+-transferrin engendered a low affinity (Kd 2 0.25 pM) for the asialoglycoprotein receptor. 1251-Fe: +-asialotransferrin was only superficially internalised by isolated hepatocytes but this was characteristic of ligands for the asialoglycoprotein receptor which have only biantennary (or single triantennary) side chains. This was also incompatible with the delivery of the ligand to sorting endosomes where the release of iron has been presumed to occur. Despite the different properties of the two ligands, the rates of iron uptake from 59Fe: +-transferrin and 59Fe: +-asialotransferrin were identical, suggesting a common mechanism for the translocation of iron across the plasma (or possibly endosomal) membrane such as a transmembrane oxidoreductase. Competition studies with unlabelled ligand or impermeable ferric ammonium citrate gave an IC50 of 1 -15 pg Fe3' /ml for this process. The absence of transferrin receptor from the surface of the terminally differentiated, quiescent hepatocyte would be compatible with the dual roles suggested for transferrin as an iron transport protein and as a growth factor. The release of iron at the hepatocyte cell surface would effectively uncouple the two functions and render the hepatocyte unresponsive to growth stimulation by transferrin.A healthy liver contains about 30% of total body iron which is stored in hepatocytes as a complex with ferritin. The net retention or release of iron from this store is determined by the serum iron concentration, which is set by the level of transferrin saturation. This in turn is set by the balance between the synthesis in the bone marrow, degradation by the reticuloendothelial system and absorption of iron from the gut (see Bothwell et al., 1979). The hepatic iron store is, therefore, the buffer for the maintenance of iron haemostasis and primarily responds to variations in iron requirement by the bone marrow.Despite its physiological importance, the mechanism of iron accumulation by hepatocytes remains a paradox. Since the expression of transferrin receptors is a characteristic of cell growth and proliferation (Huebers and Finch, 1987) the terminally differentiated, quiescent hepatocyte would be expected to show only a basal level of expression or even a complete absence of this receptor. At the same time, this cell continuously accumulates and releases ir...