Background: Cognitive impairment is prevalent in hematologic malignancy patients treated with BMT (autologous: 18.7%; allogeneic: 35.7%; Sharafeldin JCO; 2018). Given the inter-individual variability in risk of cognitive impairment in this population, we investigated the role of genetic susceptibility using a genome-wide single nucleotide polymorphism (SNP) array platform to identify novel genetic associations. Methods:Discovery: Cognitive function was assessed objectively in 239 adult BMT recipients at pre-specified timepoints: pre-BMT and at 6 mo, 1y, 2y, and 3y post-BMT. A global deficit score (GDS - a summary score of 14 standardized neuropsychological tests) was computed for each patient; a higher score indicated greater cognitive impairment. SNPs passing standard quality control filters ( > 1.4M) were used for analysis. Linear mixed effects models used GDS as the outcome, adjusted for age, sex, BMT type, baseline cognitive reserve, and the first four principal components. We used additive, codominant, and genotype models and an adjusted genome-wide significance threshold of 1.25 x 10⁻⁸. Replication: An independent cohort of 544 BMT survivors (192 cases with self-endorsed cognitive problems and 352 controls without) was used for replication. Results:Discovery: Median age at BMT was 51.3y; primary diagnoses: 47% leukemia, 32% lymphoma, 21% multiple myeloma; 57% males; 69% non-Hispanic whites: 50% allogeneic BMT, median GDS score = 0.22 (range 0-2). Forty-four SNPs were significantly associated with increased GDS (additive model: 3 SNPs; codominant model: 20 SNPs; genotype model: 21 SNPs). Estimates ranged from increase in GDS score by 0.28 points for each additional copy of risk allele, p = 1.07 x 10⁻⁸ to increase in GDS score by 1.82 points for two copies of risk allele, p = 2.3 x 10⁻¹¹. Replication: Median age at BMT was 44y; primary diagnoses: 32% leukemia, 49% lymphoma, 19% multiple myeloma; 54% males; 80% non-Hispanic whites: 34% allogeneic BMT. Three SNPs were successfully replicated: rs116334183 resides within lncRNA-SEMA6D-2, which facilitates neuronal migration; rs13286152 86kb downstream of TLE-1, which promotes neuronal survival; and rs12486041 0.36Mb downstream from lncRNA-SPTSSB-1, which regulates sphingolipid production in neuronal axons and 0.36Mb upstream from TOMM22P6 linked to neural repair. Conclusions: In this first GWAS of cognitive impairment post-BMT, we identify 3 SNPs with plausible links to genes implicated in neuronal integrity. Functional studies are currently underway.