Carcinoembryonic antigen (CEA) preparations, from various sources were compared by radioimmunoassay. The preparations studied included four CEA standards (CEA-Roche, CEA-Montreal, CEA-City of Hope, and CEA-British) and CEA from serum and liver metastases of a patient with cancer of the colon who had an extremely high concentration of serum CEA (more than 26,000 ng/ml). The data indicate that the CEA-Roche standard differs significantly from the other three CEA standards tested, and that the serum CEA from the patient was antigenically different from currently available CEA standards as well as from the CEA obtained from the patient's own liver metastases. These antigenic differences were reflected in radioimmunoassay inhibition curves that were different and that were not affected by perchloric acid extraction of CEA. Because of the antigenic variation in the serum CEA, markedly different CEA concentrations (varying by three orders of magnitude) were measurable by two different antisera (Roche and Montreal). All the various CEA standards and samples cochromatographed on columns of Sepharose-6B, despite the large antigenic variation.We postulate that CEA consists of a family of "isoantigens" with multiple antigenic determinants. We identified a serum CEA isoantigen that was different from the currently available standards. Consequently, we believe that results of radioimmunoassays currently used for CEA measurement may not represent absolute concentrations of serum "CEA", but may reflect the binding affinity of different isoantigens to a particular polyvalent CEA antiserum.After the demonstration of a tumor-specific antigen (carcinoembryonic antigen, CEA) in adenocarcinomas of the gastrointestinal tract by Gold and Freedman (1), it was shown that immunoreactive CEA was present in the plasma of almost all patients with colon carcinoma. This suggested that serologic measurement of this antigen could be a useful diagnostic test for this malignancy (2). However, the availability of a well characterized CEA standard is an important prerequisite for a valid, generally applicable CEA assay.Coligan et al. (3) isolated CEA from several gastrointestinal tumors and found that their CEA preparation (CEA-City of Hope) was immunologically indistinguishable from CEA prepared by Gold and Freedman (CEA-Montreal) (4). The "First British CEA Standard" (CEA-British) was prepared from liver metastases arising from colon carcinoma by methods similar to those used by Gold and Freedman (4) and Coligan et al. (3), and was found (by the National Institute of Biological Standards and Control, London) to be similar to the CEA-Montreal and CEA-City of Hope by immunodiffusion and polyacrylamide gel electrophoresis, as well as in amino-acid and carbohydrate composition. Hoffmann-LaRoche (Nutley, N.J.) prepared their own CEA (CEA-Roche) and developed a standardized method and kit for routine clinical assays (5, 6). Although studies in which both Roche and Montreal reagents were used showed a high degree of correlation of positive CEA te...