arthritis (RA) were tested for RF specificity towards these IgG monoclonal anti-D antibodies the incidence and titre of reactivity towards an IgG3 monoclonal anti-D antibody was considerably greater than for a polyclonal anti-D antibody of the same Gm allotype, G3m(5). This difference was not explained by the amount of each anti-D antibody which bound to erythrocytes. Furthermore, when patients with RA were divided into groups according to their Gm phenotype, sera from a greater proportion of patients negative for the phenotype G3m(5) reacted to the G3m(5) monoclonal anti-D antibodies than sera from those patients positive for this allotype. Analysis of RF reactivities towards two IgG3 and three IgGl monoclonal anti-D antibodies, each with different Gm allotypic epitopes, indicated, however, that individual serum samples contained RFs with a spectrum of specificities; some sera appeared to react to a single set of Gm alleles, whereas others also reacted to isotypic or iso-allotypic epitopes, or both. Our data suggest that RFs with specificity for Gm allotypes do not arise in patients who carry that particular allotype owing to tolerance induced in fetal-neonatal life. Conversely, RFs with apparent specificity for a Gm allotype formed in patients negative for that allotype may be reacting to a closely related but different epitope. Final proof requires precise specificities for each RF formed, and IgG3 monoclonal anti-D antibodies would be useful reagents for this purpose.