We report graft-versus-host disease (GVHD)-free relapse-free survival (GRFS) (a composite end point of survival without grade III-IV acute GVHD [aGVHD], systemic therapy-requiring chronic GVHD [cGVHD], or relapse) and cGVHD-free relapse-free survival (CRFS) among pediatric patients with acute leukemia (n = 1613) who underwent transplantation with 1 antigen-mismatched (7/8) bone marrow (BM; n = 172) or umbilical cord blood (UCB; n = 1441). Multivariate analysis was performed using Cox proportional hazards models. To account for multiple testing, P < .01 for the donor/graft variable was considered statistically significant. Clinical characteristics were similar between UCB and 7/8 BM recipients, because most had acute lymphoblastic leukemia (62%), 64% received total body irradiation-based conditioning, and 60% received anti-thymocyte globulin or alemtuzumab. Methotrexate-based GVHD prophylaxis was more common with 7/8 BM (79%) than with UCB (15%), in which mycophenolate mofetil was commonly used. The univariate estimates of GRFS and CRFS were 22% (95% confidence interval [CI], 16-29) and 27% (95% CI, 20-34), respectively, with 7/8 BM and 33% (95% CI, 31-36) and 38% (95% CI, 35-40), respectively, with UCB (P < .001). In multivariate analysis, 7/8 BM vs UCB had similar GRFS (hazard ratio [HR], 1.12; 95% CI, 0.87-1.45; P = .39), CRFS (HR, 1.06; 95% CI, 0.82-1.38; P = .66), overall survival (HR, 1.07; 95% CI, 0.80-1.44; P = .66), and relapse (HR, 1.44; 95% CI, 1.03-2.02; P = .03). However, the 7/8 BM group had a significantly higher risk for grade III-IV aGVHD (HR, 1.70; 95% CI, 1.16-2.48; P = .006) compared with the UCB group. UCB and 7/8 BM groups had similar outcomes, as measured by GRFS and CRFS. However, given the higher risk for grade III-IV aGVHD, UCB might be preferred for patients lacking matched donors.