The difficulty in distinguishing lymphocytes from other cells in the myocardial interstitium on hematoxylin-eosin-stained sections has contributed to the variable incidence of myocarditis diagnosed by endomyocardial biopsy. We investigated the value of immunoperoxidase staining using antibodies to leukocyte common antigen (LCA) for identifying interstitial mononuclear cells in formalin-fixed, paraffin-embedded, endomyocardial biopsy samples of the right ventricle obtained by bioptome from 50 consecutive autopsies. In 42 cases with nonspecific findings or no disease, two observers independently quantitated interstitial lymphocytes on hematoxylin-eosin-stained sections and LCA-positive cells on immunoperoxidase sections. Mean values for lymphocytes on hematoxylin-eosin-stained sections were significantly different for the two observers (5.8/mm2 vs. 8.9/mm2, P less than 0.001). In contrast, mean values for LCA-positive cells on immunoperoxidase sections were similar (13.8/mm2 vs. 15.1/mm2, P = not significant). Interobserver concordance was excellent for LCA-immunoperoxidase sections (r = 0.83), exceeding that obtained for hematoxylin-eosin-stained sections (r = 0.63). Intraobserver correlations between lymphocytes identified on hematoxylin-eosin-stained sections and LCA-positive cells on immunoperoxidase sections were poor (r = 0.28 and 0.14). LCA-immunoreactive cells included, in addition to lymphocytes, mast cells and histiocytes. We conclude that there is considerable interobserver discordance in identifying lymphocytes on hematoxylin-eosin-stained sections of endomyocardial biopsy samples. While LCA-immunoperoxidase staining reduces interobserver variability, results must be interpreted with caution since this antibody stains other leukocytes in addition to lymphocytes.