Detection of minimal residual disease (MRD) pre- and post-hematopoietic cell transplantation (HCT) for pediatric acute lymphoblastic leukemia (ALL) has been associated with relapse and poor survival. Published studies have had insufficient numbers to: (1) compare the prognostic value of pre-HCT and post-HCT MRD; (2) determine clinical factors post-HCT associated with better outcomes in MRD+ patients; and (3) use MRD and other clinical factors to develop and validate a prognostic model for relapse in pediatric patients with ALL who undergo allogeneic HCT. To address these issues, we assembled an international database including sibling (n = 191), unrelated (n = 259), mismatched (n = 56), and cord blood (n = 110) grafts given after myeloablative conditioning. Although high and very high MRD pre-HCT were significant predictors in univariate analysis, with bivariate analysis using MRD pre-HCT and post-HCT, MRD pre-HCT at any level was less predictive than even low-level MRD post-HCT. Patients with MRD pre-HCT must become MRD low/negative at 1 to 2 months and negative within 3 to 6 months after HCT for successful therapy. Factors associated with improved outcome of patients with detectable MRD post-HCT included acute graft-versus-host disease. We derived a risk score with an MRD cohort from Europe, North America, and Australia using negative predictive characteristics (late disease status, non-total body irradiation regimen, and MRD [high, very high]) defining good, intermediate, and poor risk groups with 2-year cumulative incidences of relapse of 21%, 38%, and 47%, respectively. We validated the score in a second, more contemporaneous cohort and noted 2-year cumulative incidences of relapse of 13%, 26%, and 47% (P < .001) for the defined risk groups.