CD59-antigen expression on the surface membranes of erythrocytes, granulocytes, monocytes, lymphocytes, and platelets was determined by flow cytometry in 34 healthy controls and 17 patients with paroxysmal nocturnal hemoglobinuria (PNH). In all PNH patients, CD59-negative erythrocytes accounted for > 10% of the total erythrocyte population. Two erythrocyte populations (CD59-negative and normally positlve or CD59-negative and partly positive), three populations (CD59-negative, partly positive, and normally positive), and one population (CD59-negative) were demonstrated in ten, six, and one patients, respectively. However, CD59-negative granulocytes did not account for > 10% of the total granulocytes in two patients, and one of them had only a CD59 normally positive granulocyte population. A particular granulocyte population extended over both CD59-negative and partly positive areas was shown in two patients. Two populations (CD59-negative and normally positive) and one population (CD59-negatlve) were demonstrated in monocytes and lymphocytes. CD59-negative lymphocytes accounted for >50% of the total lymphocytes in only two patients. Three patients had a CD59 normally positive lymphocyte population. Percentages of CD59-positive platelet population in normal controls were widely various. Therefore, it was usually difficult to discriminate between PNH-affected and normal platelets. Thus, the flow cytometric profiles of CD59-antigen expression varied not only between PNH patients but between cell lineages. The present results and our prior study indicate that CD59 flow cytometry using erythrocytes and granulocytes is most suitable for diagnosing PNH. © 1994 Wiley-Liss, Inc.