Abstract During double umbilical cord blood transplant (DUCBT), the winning unit (WU) rejects the losing unit (LU) because of WU T cells directed against the LU mismatched HLA. This immune response might protect against relapse, especially when the patient and LU (PT-LU) share the same mismatch with the WU. To validate this hypothesis, a retrospective Eurocord study, conducted on 383 DUCBTs, focused on posttransplant relapse and HLA mismatches between PT-LU and the WU. A PT-LU HLA-A mismatch shared with the WU was associated with a lower 7-year relapse incidence (16% vs 28%; P = .048). In addition, multiple PT-LU shared HLA mismatches were also associated with a lower relapse risk (7% vs 29%; P = .003). In DUCBT with ≥2 HLA mismatches between patient and WU, the number of HLA mismatches between those did not significantly affect relapse incidence, whereas multiple PT-LU shared HLA mismatches remained associated with a lower relapse risk (7% vs 29%; P = .0038). Finally, considering patients who did not develop either grade 2 to 4 acute graft-versus-host disease or chronic graft-versus-host disease, a PT-LU shared HLA-A mismatch as well as multiple PT-LU HLA mismatches shared with the WU remained associated with a significantly lower 7-year relapse incidence. In multivariate adjusted analyses multiple PT-LU shared HLA mismatches remained associated with a significantly reduced 7-year posttransplant relapse risk. Our analysis indicates that, during DUCBT, PT-LU shared HLA mismatches prime an immune response of the WU against leukemia, reducing the long-term risk of posttransplant relapse, and that DCBT has particular utility in those with high-risk leukemia.