PURPOSE There is an unmet need for therapeutic options that prolong survival for patients with heavily pretreated, metastatic castration-resistant prostate cancer (mCRPC). The phase III, open-label KEYLYNK-010 study evaluated pembrolizumab plus olaparib versus a next-generation hormonal agent (NHA) for biomarker-unselected, previously treated mCRPC. METHODS Eligible participants had mCRPC that progressed on or after abiraterone or enzalutamide (but not both) and docetaxel. Participants were randomly assigned (2:1) to pembrolizumab plus olaparib or NHA (abiraterone or enzalutamide). The dual primary end points were radiographic progression-free survival (rPFS) by blinded independent central review per Prostate Cancer Working Group–modified RECIST 1.1 and overall survival (OS). Time to first subsequent therapy (TFST) was a key secondary end point. Safety and objective response rate (ORR) were secondary end points. RESULTS Between May 30, 2019, and July 16, 2021, 529 participants were randomly assigned to pembrolizumab plus olaparib and 264 to NHA. At final rPFS analysis, median rPFS was 4.4 months (95% CI, 4.2 to 6.0) with pembrolizumab plus olaparib and 4.2 months (95% CI, 4.0 to 6.1) with NHA (hazard ratio [HR], 1.02; 95% CI, 0.82 to 1.25; P = .55). At final OS analysis, median OS was 15.8 months (95% CI, 14.6 to 17.0) and 14.6 months (95% CI, 12.6 to 17.3), respectively (HR, 0.94; 95% CI, 0.77 to 1.14; P = .26). At final TFST analysis, median TFST was 7.2 months (95% CI, 6.7 to 8.1) versus 5.7 months (95% CI, 5.0 to 7.1), respectively (HR, 0.86; 95% CI, 0.71 to 1.03). ORR was higher with pembrolizumab plus olaparib versus NHA (16.8% v 5.9%). Grade ≥3 treatment-related adverse events occurred in 34.6% and 9.0% of participants, respectively. CONCLUSION Pembrolizumab plus olaparib did not significantly improve rPFS or OS versus NHA in participants with biomarker-unselected, heavily pretreated mCRPC. The study was stopped for futility. No new safety signals occurred.