There has been recent interest in eosinophils as a histological diagnostic marker of liver allograft rejection. However, the reliability of counting eosinophils in sections stained with hematoxylin and eosin (H and E) has not been evaluated previously. We quantified eosinophils in 10 day-5 protocol liver biopsy specimens in 10 patients. The control groups were 5 patients with cytomegalovirus infection, 5 patients with obscure liver dysfunction, and 5 patients with HCV infection. Eosinophil count was assessed using H and E and by specific monoclonal antibody staining using (1) an anti-ECP antibody (EG2) and (2) a monoclonal antibody against human eosinophil major basic protein (MBP) (BMK-13). The average percentage of the total inflammatory infiltrate of eosinophils in portal tracts was 9% in the moderate to severe rejection group as compared with 0.25% in the mild rejection group (P < .001) and 0% in the control group (P < .001). The eosinophil count decreased markedly after successful treatment of rejection. The H and E staining correlated with MBP + the (BMK-13 immunoreactive) cells but were more numerous with BMK-13. BMK-13 also stained significantly more cells when compared with EG2 (P < .01). This difference may be because EG2 staining only activated eosinophils, whereas BMK-13 is a pan-eosinophilic maker, regardless of activation. This study confirms that eosinophils are a specific feature of acute cellular rejection and are an aid to its diagnosis. BMK-13 is a useful pan-eosinophilic marker that is more efficient in obtaining eosinophil count when compared with H and E.