Background: In ARIEL3, rucaparib significantly improved progression-free survival (PFS) vs placebo in all randomized patients, including patients with BRCA-mutant, BRCA wild-type/high-LOH (prespecified as ≥16% genomic LOH), or BRCA wild-type/low-LOH ( < 16% genomic LOH) rOC (Coleman et al. Lancet. 2017;390:1949-61). This exploratory analysis evaluated the optimal cutoff for percentage of (%) genomic LOH in BRCA wild-type rOC in ARIEL3. Methods: Genomic LOH of archival tumor tissue DNA was centrally assessed using Foundation Medicine’s next-generation sequencing-based assay (Cambridge, MA, USA). Treatment effect for investigator-assessed PFS (invPFS) was analyzed in BRCA wild-type rOC for the prespecified cutoff (16%) and across a range of cutoffs for % genomic LOH (5%–30%). Hazard ratios (HRs) were estimated using a stratified Cox proportional hazards model. Prognostic and predictive utility of % genomic LOH was assessed by comparing invPFS between and within the treatment arms. Results: In ARIEL3, 564 patients were randomized. Of the 368 patients with BRCA wild-type associated rOC, LOH was calculable for 319. Rucaparib significantly improved invPFS vs placebo between the 5% and 23% cutoffs for % genomic LOH. Using the prespecified cutoff (16%), the HR (rucaparib vs placebo) was 0.44 (95% confidence interval [CI], 0.29‒0.66; P< 0.0001) for patients with high-LOH rOC. To further assess the % genomic LOH cutoff, we compared patients with high- vs low-LOH rOC within the rucaparib arm and found a statistically significant benefit for invPFS at the prespecified cutoff of 16% (HR, 0.70; 95% CI, 0.50‒0.97; P= 0.0338). In the placebo arm, no statistically significant benefit was observed for invPFS in patients with high- vs low-LOH rOC at any cutoff tested. Conclusions: Rucaparib improved invPFS vs placebo across the range of cutoffs tested for % genomic LOH, including the prespecified cutoff of 16% for high LOH. The observance of significant differences between patients with high- vs low-LOH rOC in the rucaparib but not placebo arm suggests that genomic LOH is a predictive but likely not prognostic biomarker. Clinical trial information: NCT01968213