Background: IMbassador250 was a prospective phase III international trial which showed no overall survival (OS) benefit for adding atezolizumab to enzalutamide for men with mCRPC who had prior progression on abiraterone. We hypothesized that genomic biomarkers in ctDNA may identify patients who have poorer OS with 2^nd generation novel hormonal therapy. Methods: Pre-treatment (but post-abiraterone progression) plasma samples from IMbassador250 were submitted for comprehensive genomic profiling using FoundationOne Liquid CDx. We detected elevated ctDNA tumor fraction (TF) using a novel algorithm that incorporates aneuploidy as well as tumor-derived short variant signal. A pre-specified TF cutoff of ≥ 2% was defined as high. A prospective-retrospective biomarker statistical analysis plan was developed in accordance with Simon Criteria, pre-specifying analyses, cutoffs, and power assessments. Results: 494 baseline plasma specimens were evaluable. The TF high group [371 (74%)] had significantly shorter OS than TF low (median 11.0 vs. 22.1 months, HR 4.3, 95% CI 3.0 – 6.1, p < 0.0001) (primary endpoint). Pre-specified supportive analyses compared the performance to prognosticate OS (concordance [std error]) of baseline PSA alone (0.65 [0.63 – 0.67]) to a model consisting of all evaluable baseline features including PSA, treatment arm, age, race, ECOG score, hemoglobin, alkaline phosphatase, albumin, number of metastatic sites, and sites of metastasis (0.71 [0.70 – 0.73]) to TF alone (0.72 [0.71 – 0.73]) to all features plus TF (0.76 [0.74 – 0.77]). Likelihood ratio test for improvement of OS prediction by adding TF to all available clinical features: p < 0.0001. Focusing on the patients with TF ≥ 2% (n = 371) exploratory analysis observed less favorable outcomes with detection of AR amplifications (n = 203 [55%], HR: 1.4, 95%CI: 1.1 – 1.8), while individual AR mutations were not as strongly associated with OS: L702 ([16%], HR: 1.2, 95%CI: 0.88 – 1.7), W742 ([4.3%], HR: 0.59, 95%CI: 0.30 – 1.2), H875 ([9.4%], HR: 0.71, 95%CI: 0.46 – 1.1), T878 ([25.3%], HR: 0.86, 95%CI: 0.65 – 1.1). Conclusions: In this cohort TF < 2% at baseline is strongly associated with favorable OS on enzalutamide after progression on abiraterone compared to TF ≥ 2%. TF alone has comparable discriminatory ability to anticipate overall survival compared to all available clinical features combined, and adds significant prognostic power when combined with clinical features in this setting. AR amplifications, but not mutations, further improve strengths of association. Clinical trial information: NCT03016312.