Abstract Background. There is no hierarchical algorithm that weights the characteristics of individual donors against each other in a quantitative manner to facilitate donor selection when multiple potential equally HLA-matched unrelated donors (URD) are available. Donor factors, such as age, sex, CMV status, ABO type, and matching of secondary HLA loci (DQB1, DPB1), have been associated with recipient survival in URD hematopoietic cell transplantation (HCT) although the impact of specific factors has varied among studies. The goal of this study was to develop and validate a donor selection score that prioritizes donor characteristics associated with better survival in 8/8 HLA-matched URD transplantation. Methods. Two large CIBMTR patient datasets were studied: HCT from 1999-2011 (n=5952) and 2012-2014 (n=4510). Patients were adults (>18), transplanted for acute myelogenous leukemia (AML), acute lymphocytic leukemia (ALL), chronic myelogenous leukemia (CML), or myelodysplastic syndrome (MDS). Each dataset was randomly split for the analysis. Cohort 1 (c1): 2/3 (n=3969) for modeling/score development (training) and 1/3 (n=1983) for testing and similarly for cohort 2 (c2): 2/3 (n=3051) and 1/3 (n=1459). Thus, two independent models were built and tested, adjusting for significant patient characteristics associated with survival. Interactions between donor characteristics, and donor and recipient characteristics were tested. The following donor characteristics were considered for the donor score: HLA-DQB1 matching, HLA-DPB1 matching (using the T-cell epitope matching categorization), age, sex matching, parity, CMV matching, ABO matching and race matching. Results. In the final survival model (training set from 1999-2011, c1) we found significant negative associations with survival for three donor risk factors: non-permissive DPB1 matching (HR 1.13; 95% CI 1.01, 1.26; p-value=0.032), older donor age (as a linear effect, HR 1.07 per decade increase in age; 95% CI 1.02, 1.12, p-value=0.004), and CMV mismatching for CMV+ recipients (HR 1.14; 95% CI 1.02, 1.27; p-value=0.022). For CMV- recipients, a CMV+ donor was not significantly associated with an increase in mortality (HR=1.03; 95% CI 0.89-1.20; p-value=0.68), so this was not included in the score. ABO mismatching (any type: major, minor or bidirectional) was associated with mortality in initial modelling, but the effect was not present in more recent transplants (HR for ABO mismatch among patients transplanted since 2007: 1.04; 95% CI 0.91-1.19; p-value=0.638), so it was not included in the final model and donor score. Based on these results a donor risk score was constructed, however this score was not validated in the testing set (c1), nor were any of the individual component donor factors significantly associated with worse overall survival. In the second cohort (c2), only donor age was significantly associated with worse survival, and it validated in the independent test set from c2. Since donor age was significant in 3 of the 4 cohorts, we quantified the impact of donor age in the validation set of the most recent cohort, c2. We found that choosing a donor 2, 5, 10 or 20 years older was associated with a 1%, 2%, 3% or 7% decrease in 2 year OS, adjusted for patient characteristics. Conclusion. Despite data on over 10,000 URD transplants, we were unable to develop a valid donor selection score. The only donor characteristic associated with better survival was younger age, with 2-year survival being 3% better when a donor is 10 years younger. We did not test other endpoints; it is possible that separate scores could be generated to predict the risk of other outcomes (e.g. graft failure, graft-versus-host disease), however, unless the adverse donor characteristics are identical for these outcomes, centers will still have to prioritize the various donor characteristics to select from a pool of potential donors. This large data set shows that none of the other easily available donor clinical and genetic factors tested were reproducibly associated with survival and hence, flexibility in selecting URD based on these characteristics is justified. These data support a simplified URD selection process and have significant implications for URD registries. Disclosures Porter: Incyte: Honoraria; Genentech/Roche: Employment, Other: Family member employment, stock ownship - family member; Servier: Honoraria, Other: Travel reimbursement; Novartis: Honoraria, Patents & Royalties, Research Funding; Immunovative Therapies: Other: Member DSMB. Lee: Amgen: Other: One-time advisory board member; Bristol-Myers-Squibb: Other: One-time advisory board member; Mallinckrodt: Honoraria; Kadmon: Other: One-time advisory board member.