In this study, we investigated the effects of low-dose antithymocyte globulin (ATG, thymoglobulin) in the prevention of acute graft-versus-host disease (aGVHD) in mismatched, unrelated hematopoietic stem cell transplantations (uHSCTs) in patients with the single disease entity of acute myelogenous leukemia (AML). Patients (n = 103) with a variable risk for AML who received uHSCTs from available Asian and Caucasian donors were enrolled. First, we compared HLA-matched (group 1, n = 54) and HLA-mismatched (group 2, n = 49) transplantation patients. Then, we divided the patients in group 2, who had received transplants from allele(s)/antigen-mismatched donors, into 2 subgroups: patients who used ATG (group 3, n = 24) and those who did not (group 4, n = 25). To prevent the development of aGVHD, the patients in group 3 received ATG at a dose of 1.25 mg/kg body weight per day for 2 consecutive days, together with our standard regimen of methotrexate (MTX) and tacrolimus. The median CD34(+) cell infusion was 4.2 x 10(6)/kg (range: 1.2-34.4). The median patient age was 41 years (range: 16-57), and the median follow-up duration of patients who were event-free survivors was 23 months (range: 2-72). The overall incidences of aGVHD and chronic GVHD (cGVHD) were 38% and 56%, respectively. Of 48 evaluable patients in group 2, 10 (21%) developed moderate to severe aGVHD (grades II-IV). In contrast, 2 (8%) of the 24 patients in group 3 and 7 (29%) of the 24 evaluated patients in group 4 required therapy for aGVHD (grades II-IV; P = .038). The incidence of cGVHD was not different between groups 3 and 4. The estimated probabilities of overall survival (OS) and event-free survival (EFS) at 2 years for group 2 were 55% and 44%, respectively. In comparison, the estimated probabilities of OS and EFS at 2 years for groups 3 and 4 were 68% versus 38% (P = .043) and 58% versus 38% (P = .103), respectively. The overall cumulative incidence of nonrelapse mortality (NRM) was 29% in group 2. The cumulative incidence of NRM differed markedly between group 3 (16%; 95% confidence interval [CI], 4%-28%) and group 4 (44%, 95% CI, 34%-54%) (P = .013). We found no difference in cytomegalovirus (CMV) reactivation between groups 3 and 4. These results suggest that in mismatched uHSCT, a low dose of ATG (total 2.5 mg/kg) may prevent moderate to severe aGVHD, with comparable rates of relapse and CMV reactivation and a greatly decreased rate of NRM.