We studied the analytical and clinical performance of six methods for creatine kinase (EC 2.7.3.2) isoenzyme MB (CK-MB): three immunoassays (Behring, Hybritech, and International Immunoassay Labs); one immunoinhibition assay (Roche); one immunoinhibition/column method (Du Pont); and one electrophoretic method (Beckman). Between-day precision for all kits was poor at the upper reference limit. All methods gave results linearly related to CK-MB concentration and all were free from CK-MM, CK-BB, and adenylate kinase interference. Only the Du Pont method was adversely affected by atypical isoenzymes. For diagnosis of acute myocardial infarction in a coronary care population (n = 40; prevalence = 45%), all methods were approximately 95% efficient, when appropriate reference criteria were used. Some manufacturers fail to provide data for an appropriate (acutely ill, non-infarct) reference population; decreased diagnostic specificity may result from use of reference ranges based on results for healthy subjects. Expression of CK-MB as a percent of total CK degrades efficiency unless total CK is markedly increased.