This is a multicenter retrospective comparison of 2 myeloablative conditioning regimens in 454 patients with acute myeloid leukemia (AML) in remission: busulfan (4 days) and fludarabine (BUFLU) versus thiotepa, busulfan, and fludarabine (TBF). Eligible for this study were patients allografted between January 2008 and December 2018 in 10 transplant centers, with AML in first or second remission: 201 patients received BUFLU, whereas 253 received TBF. The 2 groups (BUFLU and TBF) were comparable for age (P = .13) and adverse AML risk factors (P = .3). The TBF group had more second remissions and more haploidentical grafts. The donor type included HLA-identical siblings, unrelated donors, and family haploidentical donors. The 5-year cumulative incidence of nonrelapse mortality (NRM) was 19% for BUFLU and 22% for TBF (P = .8), and the 5-year cumulative incidence of relapse was 30% and 15%, respectively (P = .0004). The 5-year actuarial survival was 51% for BUFLU and 68% for TBF (P = .002). In a multivariate Cox analysis, after correcting for confounding factors, the use of TBF reduced the risk of relapse compared with BUFLU (P = .03) and the risk of death (P = .03). In a matched pair analysis of 108 BUFLU patients matched with 108 TBF patients, with the exclusion of haploidentical grafts, TBF reduced the risk of relapse (P = .006) and there was a trend for improved survival (P = .07). Superior survival of patients receiving TBF as compared with BUFLU is due to a reduced risk of relapse, with comparable NRM. The survival advantage is independent of donor type and AML risk factors.