Fifty-four anti-MUC1 antibodies submitted to the International Society for Oncodevelopmental Biology and Medicine (ISOBM) Workshop (TD-4) were evaluated in immunoradiometric assays, using sera from carcinoma patients and healthy donors. The carcinoma serum pool contained sera from 30 patients with advanced cancer (10 breast, 10 colon, and 10 ovarian). This serum pool contained 696 kU/l MUC1, 770 micrograms/l CEA, and 3,700 kU/l CA 125. The reference serum pool was obtained from 10 healthy women combined with 20 sera from pregnant women, of which half had elevated CA 125 (range 82-254 kU/l). The reference serum pool contained 13 kU/l MUC1, 2 micrograms/l CEA, and 65 kU/l CA 125. The Workshop antibodies were tested both as solid-phase antibodies and as tracer antibodies with the carcinoma serum pool. Twenty-two tracer antibodies and 38 solid-phase antibodies gave at least one combination with > 10% binding of the tracer antibody for a total of 836 combinations. These were tested further with the reference serum pool. Antibodies used as tracers could be separated into three categories: Group 1 antibodies, MF06, MF11, B27.29, MF30, and Ma552, gave mainly 'carcinoma-specific' assays in combinations with the solid-phase antibodies, i.e. binding ratio between carcinoma MUC1 and reference MUC1 > 10. Group 2 antibodies, DF3, 7540MR, A76-A/C7, BC4N154, M38, 7539MR, B12, GP1.4, 232A1, Mc5, and Ma695 gave both 'specific' and 'nonspecific' binding ratios depending on the solid-phase antibody used. Group 3 antibodies, 214D4, BC4E549, E29, BCP8, BC3, and 3E1.2, gave mainly 'nonspecific' combinations, i.e. ratios < or = 10. All antibodies used to capture MUC1 on the solid phase gave both 'specific' and 'nonspecific' combinations depending on the tracer antibody used. Ten antibodies were clearly more efficient as solid-phase capture antibodies; Ma695, B12, M38, GP1.4, 214D4, MF06, B27.29, A76-A/C7, BC3, and KC4. Our findings indicate that the ability to detect 'carcinoma-specific MUC1' cannot be deduced from epitope specificity alone.