The epitope specificities of 30 monoclonal antibodies (MAbs) against the most common human cytokeratins. i.e., Nos. 8, 18, and 19, in epithelial cells were investigated in the ISOBM TD-5 Workshop. Seven research groups from universities or companies participated independently in the evaluation of the antibody specificities. The complex assembly of cytokeratins in vivo, with obligatory heterologous dimeric combinations of different cytokeratins from each of the two major groups, comprising together more than 20 different individual cytokeratins, made analysis of the antibody reactivity patterns with isolated single cytokeratins necessary. The concordance of the evaluations was striking and independent of the technologies used. As antigens purified individual cytokeratins, chemically degraded purified cytokeratins, recombinant intact and truncated cytokeratins, as well as specific synthesized shorter peptides were used. In order to elucidate the epitope specificity, reactivity patterns in ELISA assays and immunoblots with partial enzymatic degradation of the antigens were performed. Competitive cross-inhibition experiments between antibodies using antigens and antibodies in all possible combinations were performed with radioimmunometric assays, BIAcore, and ELISA technology. All 30 antibodies could convincingly be classified with regard to target cytokeratin. One MAb (192) had to be deleted due to dual specificities in both isotype and epitope specificity against its target. Six antibodies bound selectively to cytokeratin 8, 14 to cytokeratin 18, and 10 to cytokeratin 19, as demonstrated by using native, recombinant, and synthesized antigens. The immunodominant part of the molecule for all three types of cytokeratins was located in the region of amino acid (aa) 270-400. Out of the six MAbs reactive with cytokeratin 8, four MAbs, i.e., 178, 199, 202, and 206, were reactive with a sequence in the interval aa 340-365, and MAb 191 reacted with a closely related epitope. The remaining antibody, 192, presented dual specificities. At least two closely related major immunogenic epitopes could be identified in cytokeratin 8. In cytokeratin 18 four distinct epitopes could be documented, again with the dominating sequence region 270-429 as target for 10 (181, 184, 186, 188, 189, 190, 193, 196, 198, and 200) out of 14 antibodies. Since MAb 193 is known to react with the M3 epitope, aa 322-342 in cytokeratin 18, this entire group is reactive in the region close to the charge shift, in the middle of the rod 2B region, as shown by competitive binding. The remaining four anticytokeratin 18 antibodies (180, 185, 203, and 205) displayed unique, noncompetitive binding to this filament. Cytokeratin 19, reactive with altogether ten antibodies, displayed two major epitopes, all of them also within the large immunodominant region. MAbs 179, 195, 197, and 204 were reactive with the peptides aa 311-335 also known as the KS 19.1 epitope, and MAbs 182, 183, 187, 194, and 201 bound to peptide aa 346-367, known as the BM 19.21 epitope. One antibody, 231, was selectively reactive with aa 356-370 in cytokeratin 19. A complex pattern of binding specificities comprising at least ten different, noncompetitive epitopes, mainly situated in the rod portion, 2A and 2B, situated close to the charge shift in the rod of all three cytokeratins was documented. Out of the 29 classifiable antibodies, altogether 22 were reactive in this very short region, i.e., from aa 311 to 370 in all cytokeratin filaments. The remaining seven antibodies displayed unique binding properties. The implications of the findings are of significance both for immunohistochemistry and for assaying circulating heterodimeric, partially degraded complexes in patients' blood for tumor marker evaluation.