Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an effective therapy for patients with relapsed acute myeloid leukemia. In this retrospective, multicenter study, we analyzed the outcome of 63 children (median age, 7 y; range, 0.2 to 17) who received unmanipulated allo-HSCT in second complete remission. Either a matched family donor or an unrelated donor was used in 29 (46%) and 34 (54%) patients, respectively. The stem cell source was bone marrow in 53 children (84%), peripheral blood in 7 (11%), and cord blood in 3 patients (5%). Preparative regimen included total body irradiation in 25 patients (40%). The 5-year estimates of overall survival and leukemia-free survival were 53% [95% confidence interval (CI) 39-66] and 49% (95% CI 35-63), respectively, whereas the cumulative incidence of relapse and transplant-related mortality (TRM) were 26% (95% CI 16-41) and 25% (95% CI 15-40), respectively. In multivariate analysis, the use of a matched family donor predicted a better probability of LFS [relative risk (RR) 2.29, P=0.05]. Both chronic graft-versus-host disease occurrence and age at diagnosis greater than 11 years were associated with an increased TRM (RR 8.08, P=0.04 and RR 4.38, P=0.05, respectively). These results indicate that allo-HSCT is a procedure able to rescue a significant proportion of children with acute myeloid leukemia in second complete remission, especially if an human leukocyte antigen-compatible relative is employed as donor. Both leukemia recurrence and TRM contributed to treatment failure. Optimization of donor selection and of strategies for both prophylaxis and treatment of graft-versus-host disease may improve the results of unrelated donor allo-HSCT.