Refractory cytopenia of childhood is the most common subtype of myelodysplastic syndrome in children. In this study, we compared the outcome of immunosuppressive therapy using horse antithymocyte globulin (n=46) with that using rabbit antithymocyte globulin (n=49) in 95 patients with refractory cytopenia of childhood and hypocellular bone marrow. The response rate at 6 months was 74% for horse antithymocyte globulin and 53% for rabbit antithymocyte globulin (P=0.04). The inferior response in the rabbit antithymocyte globulin group resulted in lower 4-year transplantation-free (69% versus 46%; P=0.003) and failure-free (58% versus 48%; P=0.04) survival rates in this group compared with those in the horse antithymocyte globulin group. However, because of successful second-line hematopoietic stem cell transplantation, overall survival was comparable between groups (91% versus 85%; P=ns). The cumulative incidence of relapse (15% versus 9%; P=ns) and clonal evolution (12% versus 4%; P=ns) at 4 years was comparable between groups. Our results suggest that the outcome of immunosuppressive therapy with rabbit antithymocyte globulin is inferior to that of horse antithymocyte globulin. Although immunosuppressive therapy is an effective therapy in selected patients with refractory cytopenia of childhood, the long-term risk of relapse or clonal evolution remains.