Results of an immunoenzymetric assay (TANDEM-E CKMB) for creatine kinase (CK; EC 2.7.3.2) MB isoenzyme, in which subunit-specific monoclonal antibodies are used, were compared with those by an immunochemical method (Isomune-CK) and electrophoresis (Corning agarose gel). The study involved 200 patients; greater than 500 samples were analyzed by all three methods. The analytical performances were acceptable. Between-method correlation coefficients ranged from 0.881 to 0.975. Two reference intervals were established for the immunoassays: 0-4 micrograms/L (TANDEM) and 0-4 U/L (Isomune) for "normal" patients; 0-9 micrograms/L (TANDEM) and 0-14 U/L (Isomune) for noninfarct patients. Agreement with respect to increased CK-MB as defined by the reference intervals for the noninfarct patient was 96% between TANDEM and electrophoresis, 90% between Isomune and electrophoresis. All three methods are acceptable for use in determining CK-MB, but the overall diagnostic efficiencies for the mass or activity concentration of the isoenzyme and for its proportion of total CK activity, based on the predictive value model, are 92% (electrophoresis, 0-7 U/L), 90% (electrophoresis, 0-4%), 92% (TANDEM, 0-9 micrograms/L), 88% (TANDEM, 0-3% index), 88% (Isomune, 0-14 U/L), and 83% (Isomune, 0-4%). All three methods can detect CK-MB in serum, but its presence is not necessarily diagnostic of acute infarct. We recommend using the actual concentration of CK-MB to evaluate patients with suspected acute myocardial infarct, and the percentage of CK-MB when total CK is very high.