The clinical suspicion of myocarditis relies strongly on endomyocardial biopsy for confirmation, yet the sensitivity of the procedure in this setting has not been clearly defined. Biopsy sensitivity was determined in 14 hearts with histologically proved myocarditis studied ex vivo, including 12 autopsy hearts and 2 native hearts explanted at cardiac transplantation. With use of the Stanford and Cordis bioptomes, endoymocardial biopsy was performed near the apex on the right side of the ventricular septum (four to five samples/bioptome per patient) and repeated in the nonapical portion of the septum from the moderator band to the subpulmonary infundibulum (additional three to five samples/bioptome per patient). In a casewise assessment, 43% to 57% of the endomyocardial samples were diagnostic for myocarditis, as calculated separately for each bioptome in each region of sampling (apical/nonapical). Both apical and nonapical sensitivity improved to 64% when the findings of the two bioptomes were combined (eight to nine samples/patient in each region). By collectively analyzing all available samples for each patient, 11 (79%) of 14 cases could be diagnosed, but this required a mean of 17.2 samples/patient, a number clinically unrealistic. The exclusion of four cases of fungal myocarditis from analysis did not significantly alter the results. In transmural ventricular sections, none of four patients with sudden death had inflammatory disease confined to the conduction system. In conclusion, despite six to eight negative biopsy samples/patient with suspected myocarditis, repeat biopsy may still be warranted.