In this study we retrospectively evaluated the effect and outcome of a boost dose of donor stem cells without additional chemotherapy or total body irradiation. Between March 1983 and August 1999, 20 of 788 (2.5%) patients receiving allogeneic bone marrow transplantation (BMT) were treated with an additional boost dose of donor cells. The reasons for the use of the boost treatment were primary graft failure (early rejection; n = 7), secondary graft failure including late rejection (n = 10), refractory pure red cell aplasia caused by the remaining recipient cells producing anti-erythrocyte antibodies (n = 2), and donor lymphocyte infusion induced pancytopenia (n = 1). The patients were aged from 17 to 48 years (median age 31 years). The underlying diseases of the patients were severe aplastic anemia in 12 patients, acute myelogenous leukemia in 3, acute lymphocytic leukemia in 3, and chronic myelogenous leukemia in 2. The donors were human leukocyte antigen-identical siblings in 18 cases, 1 mismatched related donor, and 1 unrelated donor. The cell source was bone marrow in 6 cases and peripheral blood progenitor cells in 14. The median interval between BMT and the boost treatment was 7 weeks (range 1–124). No conditioning regimen was given prior to the boost treatment for 11 patients, while 4 received total nodal irradiation (TNI) plus antithymocyte globulin (ATG), 3 ATG alone, and 2 TNI plus steroid. The median infused booster mononuclear cell dose was 2.55 × 108/kg (range 0.28–37.0). Fifteen (75%) patients achieved a hematological recovery. After the boost treatment, 6 of 20 (30%) patients developed acute graft-versus-host disease (GVHD) ≧ grade II, 3 of whom had had prior GVHD. Five (31.3%) of the evaluable 16 patients developed chronic GVHD. The GVHDs were easily controlled using immunosuppressive agents except in the case of 1 patient. Five patients died after the boost treatment; 2 within 30 days, 2 within 60 days, and 1 after 32 months. The causes of death were: 3 engrafment failures, 1 late rejection, and 1 infection following GVHD. With a median follow-up of 31.5 months (range 6–92), the Kaplan-Meier method estimated that the overall survival rate 1 and 3 years after the boost treatment was 80 and 71%, respectively. The survival of patients with primary graft failure was determined to be significantly lower compared to that of patients with secondary graft failure, using the log rank test (p = 0.0176). Disease category, stem cell source, conditioning prior to a boost treatment, and year of boost treatment did not have an influence on survival. We conclude that the reinfusion of donor stem cells is frequently successful in achieving engraftment with rare occurrence of fatal GVHD. Furthermore, relatively good long-term survival was demonstrated.