We compared transplant outcomes of 708 acute myeloid leukemia (AML) patients receiving haploidentical allogeneic hematopoietic-cell transplantation using thiotepa/busulfan/fludarabine (TBF) conditioning with posttransplant cyclophosphamide (ptCy), to 2083 patients receiving matched unrelated donor (MUD) transplantation using fludarabine/busulfan (FB) conditioning and in vivo T-cell depletion. For intermediate cytogenetic risk AML transplanted in first complete remission (CR1), multivariate analysis revealed that haplo-TBF significantly increased nonrelapse mortality (NRM) (HR 2.1; p = 0.0006) but did not affect relapse incidence (RI), leukemia-free survival (LFS), overall survival (OS), or graft-versus-host disease-free, relapse-free survival (GRFS). For high cytogenetic risk AML transplanted in CR1, haplo-TBF significantly increased NRM (HR = 2.7; p = 0.02), decreased RI (HR = 0.45; p = 0.03) but had no influence on LFS, OS, or GRFS. For AML transplanted in CR2, haplo-TBF significantly increased NRM (HR = 2.36; p = 0.008), decreased RI (HR = 0.38; p = 0.005), but had no influence on LFS, OS, or GRFS. Finally, for AML patients transplanted with active disease, haplo-TBF had no influence on transplant outcomes. In conclusion, compared to MUD-FB, haplo-TBF increased NRM, reduced RI in high-risk AML in CR, resulting in similar LFS, OS, and GRFS. These results comparing two different approaches support the use of a haploidentical family donor for high-risk AML patients lacking a matched sibling donor.