I should like to discuss an approach to the problem of immunological tolerance, which is closely akin to Burnet & Fenner’s (1949) ‘self-marker’ hypothesis, although its emphasis is less on the question of how antibodies are produced than on the detailed fate of potentially antigenic molecules when they leave the plasma and lymph and are catabolized. Dr Cinader’s figures certainly suggest that in tolerant rabbits heterologous albumin is eliminated from the plasma at rates which are substantially identical with those of homologous albumin. This is true also of heterologous plasma proteins in normal rabbits during the period before antibodies appear. Furthermore, we have found that in normal rats, in which albumin and globulin are eliminated at markedly different rates (Campbell, Cuthbertson, Matthews & McFarlane 1956), heterologous albumin behaves much like rat albumin and heterologous γ -globulin like rat γ -globulin for a week or more. It is a reasonable working hypothesis that foreign materials in the circulation are in the first place removed from the tissue fluids by the same mechanisms as deal with native materials possessing similar physico-chemical properties (e.g. molecular weight, surface charge). We are largely ignorant of what these mechanisms might be, except for the part played by reticulo-endothelial cells in removing particulate matter and denatured or aggregated proteins. However, the use by Coons, Leduc & Kaplan (1951) of fluorescent antibody to locate antigenic material in histological sections has shown that apparently undenatured proteins (e.g. bovine albumin, human γ -globulin) introduced into the plasma are detectable in immunologically active form within a wide variety of cells, such as renal tubule cells, liver parenchyma or vascular endothelium, in addition to cells of the reticuloendothelial system proper. There are possibly at least two processes going on concurrently: first, a non-selective ingestion of their surrounding fluid (with all that it contains) by cells which are widely distributed in the body, and are not necessarily confined to the reticulo-endothelial system. The mechanism may well be one of vacuolar ingestion (‘pinocytosis’), for which there is some circumstantial evidence, but which also presents difficulties. This would be one mechanism responsible for normal plasma protein catabolism. Secondly, there may be a selective removal of particulate or aggregated material, especially when negatively charged, by cells of the reticulo-endothelial system, notably by liver Küpffer cells.