Endomyocardial biopsy (EMB) of the right ventricle has demonstrated clinical use primarily for the detection of allograft rejection. Since its introduction, the procedure has been demonstrated to be safe and adaptable to an ambulatory population. The detection and grading of rejection have relied heavily on the criteria originally proposed by the Stanford group. Similar criteria are under study for diagnosis of myocarditis. Using autopsy-acquired cardiac allograft specimens, we report the reliability of EMB for grading of rejection and establish the dependence on the morphologic features used as the diagnostic cut-off and the fragment number obtained at biopsy. The reliability of an interpretation protocol based on the pattern of inflammation present in the multiple fragments submitted is established. We compare the reliability for rejection prediction between this approach and the currently accepted approach based on the presence of myocyte necrosis.