Among 38 hearts from autopsies in which lymphocytic myocarditis contributed to death, 10 endomyocardial specimens from the apical septal aspect of each ventricle (760 specimens) and 6 slices of ventricular myocardium (228 slices) were evaluated for myocarditis by the Dallas criteria. For each case, the number of positive biopsy samples correlated well with the mean lymphocyte counts in biopsy tissues (P less than 0.0001) and the mean number of inflammatory foci per square centimeter in myocardial slices (P less than 0.001). Right ventricular biopsy specimens, however, were positive in only 63% of the 38 cases and 17% of the 380 specimens. Similarly, left ventricular biopsy tissues were positive in only 55% of the cases and 20% of the specimens. Sampling error was somewhat more prevalent among the 11 cases with isolated myocarditis than in the 27 with myocarditis and other illnesses. Even when 10 biopsy specimens per ventricle were evaluated, the frequency of false-negative results was 45% for the left and 37% for the right ventricle. Although myocarditis was noted in 68% of the 38 septal slices, it involved the subendocardium of the right ventricle (from which biopsy specimens are usually obtained) in only 24%. Because of the mild and focal nature of the inflammatory infiltrates and involvement of regions inaccessible to the bioptome, sampling error contributes appreciably to false-negative results in endomyocardial biopsy tissue from patients with myocarditis. Thus, when myocarditis is evaluated by biopsy alone, only positive findings are considered diagnostic.