Serum antibodies to the stratum corneum of rat oesophagus epithelium, so-called 'antikeratin antibodies', have been largely demonstrated in rheumatoid arthritis (RA). IgM and IgG antibodies to this epithelium were studied by semiquantitative immunofluorescence in 528 patients with perfectly characterised rheumatic diseases, including 178 with classical or definite RA. Histological analysis of IgG antibodies showed that only antibodies which produce a linear laminated pattern restricted to the stratum corneum (IgG antikeratin antibodies) are highly specific for RA; all the other labelling patterns are not disease specific. By a semiquantitative evaluation of the stratum corneum fluorescence intensity it was shown that the diagnostic value of IgG antikeratin antibodies closely depends on their titre and it was established in objective conditions that the sensitivity is 43.26% when the specificity reaches 99.14%. A high titre of IgG antikeratin antibodies was actually pathognomonic for RA. Both the histological and semi-quantitative analyses showed that IgM antibodies to rat oesophagus epithelium, though frequently detected, are of no diagnostic value, either for RA or for any other rheumatic disease that was studied. From a review of all the international reports on IgG antikeratin antibodies it was found that, to date, 4080 patients, including 1694 with RA, have been assayed for antikeratin antibodies by 11 different research groups. Analysis of all the results obtained under comparable technical conditions showed that IgG antikeratin antibodies constitute the most specific serological criterion for the diagnosis of RA. Furthermore, it was found that their incidence does not depend on disease duration: they are present in one third of rheumatoid factor negative patients with RA, and they seem to be related to disease severity or activity, or both. Their detection in the diagnosis of rheumatic diseases should become systematic.