Background: Approximately 5-10% of men with localized prostate cancer have homologous recombination deficiency (HRD) mutations, yet it is unknown if these vary by Gleason score. In order to help design neoadjuvant trials for the HRD+ population and to minimize the number-needed-to-screen, we sought to determine if the prevalence of HRD mutations is influenced by Gleason grade. Methods: 484 localized prostate adenocarcinoma cases with somatic DNA sequencing data and Gleason grade information were interrogated using The Cancer Genome Atlas (TCGA) repository. HRD mutations were defined as pathogenic lesions in the following genes: ATM, BRCA1/2, CDK12, CHEK1/2, FANCA, FANCD2, FANCL, GEN1, NBN, PALB2, RAD51 and RAD51C. A separate analysis was also conducted restricting the gene list to BRCA1/2 and ATM, since deficiencies in these genes have the strongest association with PARP-inhibitor response. R software was used for analyses. Results: The prevalence of (any) HRD mutation was 11.6%, 10.8%, 14%, 16.8% and 25.0% for Gleason sums 6, 7, 8, 9 and 10, respectively. When analyzed using Grade Groups, the prevalence of (any) HRD mutation was 11.6%, 6%, 17.8%, 14% and 17% for Grade Groups 1, 2, 3, 4 and 5, respectively (Table). Those in Grade Group 3 and greater had 2.5 times higher odds of harboring at least one HRD mutation compared to those in Grade Groups 1 and 2 (16.8% vs 7.3%,P= 0.002). When specifically considering only BRCA1/2 or ATM lesions, the mutation prevalence was 6.1 times higher in Grade Group 3 and above compared to Grade Groups 1 and 2 (8.9% vs 1.6%, P= 0.0006) (Table). Conclusions: HRD mutations in general, and BRCA1/2/ATM mutations specifically, are enriched in localized prostate cancers with Gleason Grade Group 3 and higher (i.e. primary pattern 4 and higher). We propose that these patients could be targeted for neoadjuvant (or adjuvant) clinical trials using PARP inhibitors in HRD+ populations.