The BMT CTN 0901 phase 3 trial showed improved relapse-free survival (RFS) in patients with AML or MDS with myeloablative-conditioning (MAC) vs. Reduced intensity conditioning (RIC) allogeneic hematopoietic cell transplantation (allo-HCT). However, overall survival (OS) increase was not statistically significant because of higher non-relapse mortality (NRM) compared to RIC allo-HCT. HCT-CI ≥ 3 also predicts for poor OS related to high NRM in patients undergoing MAC CD34+ selected allo-HCT. We aimed to evaluate impact on OS, NRM and RFS of MAC CD34+ selected vs. RIC allo-HCT in patients with high comorbidity burden. We retrospectively compared the outcomes of 221 patients with AML or MDS with HCT-CI ≥ 3 who underwent MAC CD34+ selected (n=157) vs. RIC allo-HCT (n=64) at our institution from 2009 to 2017. The indication for MAC CD34+ vs. RIC allo-HCT was based on multidisciplinary consensus and primary physician's choice. Patients in the RIC group were older, had a higher proportion of high disease risk index (DRI), and higher use of matched unrelated donors compared to the MAC CD34+ group. HCTCI grades were balanced between the two cohorts, but the RIC group had a higher though non-significant proportion of patients with HCTCI > 5 (Figure 1). When comparing the MAC CD34+ vs. unmodified RIC cohorts, respectively, no difference was seen in OS 43% (95% CI 32-59) vs. 53% (95% CI 46-52), p=0.37, NRM 31% (95% CI 19-44) vs. 30% (95% CI 23-38), p=0.74, and relapse-free survival (RFS) 34% (95% CI 23-49) vs. 48% (95% CI 41-57) p=0.1, even when adjusting for age, DRI, and donor differences (Figure 2). However, incidence of acute and chronic graft versus host disease (GVHD) was lower for MAC CD34+ allo-HCT. When evaluating outcomes by HCT-CI subgroups (3, 4-5 and >5), MAC CD34+ allo-HCT resulted in higher 3-year estimated RFS 54% (95% CI 44-68) vs. 28% (95% CI 14-58), p=0.03 in patients with HCT-CI = 3, though this did not translate into an OS advantage 44% (95% CI 27-73) vs 59% (95% CI 48-72), p=0.26, possibly related to higher though non-significant NRM 16% (95% CI 4-35) vs. 28% (95% CI 18-39), p=0.26 (figure 3). When adjusting for differences in age, DRI, and donor, the difference in RFS was no longer statistically significant (p=0.13). Though limited by their retrospective nature, our data suggest that CD34+ selected allografts after myeloablative conditioning can be feasible in selected patients with comorbidity scores over 3 with low rates of acute and chronic GVHD. However, further prospective studies are needed to reduce NRM and optimize this treatment platform in older patients with comorbidities.