The introduction of immunosuppressive therapy for treatment of aplastic anemia has led to a considerable improvement in the prognosis of this disease. However, long-term follow-up of these patients showed a high incidence of "late" hematologic complications such as myelodysplasia and paroxysmal nocturnal hemoglobinuria (PNH). The detection of the glycosylphosphatitylinositol (GPI)-anchoring defect on peripheral blood cells of patients with aplastic anemia is now available as a new tool for early specific detection of PNH and is more sensitive than the Ham-test. Granulocytes appear to be the first cells affected in 11 patients with a GPI-anchoring defect of 29 suffering from aplastic anemia investigated in the present study. The later involvement of erythrocytes and a positive Ham test was observed in 1 patient. From our data it can be concluded that the rate of PNH resulting from aplastic anemia might be higher than reported in the literature when the Ham test alone was used for follow-up. Furthermore, our results suggest the clinical response to immunosuppressive therapy appears to be worse in the group developing the GPI-anchoring defect than in the group without this deficiency.