Background: Homologous recombination deficiency (HRD), defined as BRCA1/2 mutation (BRCAmut)or high genomic instability, is currently used to identify patients (pts) with epithelial ovarian cancer (EOC) most likely to benefit from PARP inhibitors. While these genomic tests are useful, they are imperfect: some BRCAm EOC demonstrate primary PARPi resistance and some HR-proficient benefit. Another approach to evaluate HRD is to measure the capacity of tumor cells to recruit nuclear RAD51 foci during S/G2 phase in the presence of double strand DNA damage using multiplexed immunofluorescence (IF) for RAD51, geminin (GMN) and yH2AX. We aimed to describe for the 1^st time HRD using this RAD51 functional assay in EOC and correlate RAD51 status to platinum response and BRCAmut. Methods: Tumor samples and clinical data were collected prospectively from pts in the randomized CHIVA trial of neoadjuvant platinum chemotherapy +/- nintedanib. IF for RAD51, GMN, and DAPI was performed on a 3uM slide from FFPE blocks, where feasible, yH2AX was positively scored on a consecutive slide. Tumors were considered RAD51-deficient if < 10% of gem+ tumor cells (TC) had > 5 RAD51+ foci. BRCAmut were identified by NGS. Results: 155 baseline chemotherapy naïve EOC samples were available. All were advanced stage (IIIC/IV), 75% were G3, 7% G2, 2% G1, and 16% grade UK. A contributive RAD51 result was obtained for 90% (139/155) of samples. Contributive NGS results were available for 130 samples. Overall, yH2AX scores were high (median % TC+: 86%, IQR: 56%-100%) confirming the presence of significant basal DNA damage in high grade EOC. Only 8 samples were yH2AX-low, including two of the three G1 tumors. In contrast, 55% (76/155) of samples were considered RAD51-deficient (score < 10%). With regard to outcome, pts with RAD51-deficient tumors had significantly higher overall response rates to neoadjuvant platinum (68% vs 37%, p = 0.04) and significantly longer median progression-free survival (HR 0.50, IC95% 0.25-0.98, p = 0.02). Considering BRCA status, 15% of tumors harbored a deleterious BRCAmut and 67% of these were RAD51-deficient. Importantly among BRCAmut EOC, the RAD51-proficient tumors had significantly poorer response to neoadjuvant chemotherapy (RR = 17% vs 75%, p = 0.02). Conclusions: We evaluated a novel functional assay of HR functionality in advanced EOC. The assay requires minimal tissue and yields contributive results in 90% of cases. Overall, EOC demonstrate high levels of basal DNA damage, yet 55% fail to recruit RAD51 foci during S/G2 cell cycle phase. These RAD51-deficient EOC have improved outcome after neoadjuvant platinum. Conversely, the RAD51 assay also identified a small subset of RAD51-high BRCAmut tumors with poor platinum response. Whether this RAD51 functional assay may also predict PARP inhibitor benefit is currently being investigated.