Background: A small proportion of metastatic PC exhibit outlier somatic mutation rates. The incidence, clinical course and treatment response of pts with hypermutation (HM) is poorly characterised. Methods: We performed targeted sequencing of 1047 plasma cell-free DNA samples and calculated somatic mutation burden. HM samples and available matched archival tissue were additionally subjected to whole exome sequencing. Trinucleotide mutational signatures and microsatellite instability (MSI) were determined via nonnegative matrix factorization and mSINGS, respectively. We evaluated PSA decline ≥50% from baseline (PSA50), time from androgen deprivation therapy (ADT) to castration-resistant prostate cancer (CRPC), median duration of 1^st line CRPC therapy (1L CRPCT) and median OS (time from CRPC to death). The control cohort consisted of 199 CRPC pts treated with 1L abiraterone + prednisone (ABI+P) or enzalutamide (ENZ). Results: 671 samples from 434 pts had ctDNA% > 2 and were evaluable. The median mutation rate was 2.59/Mb (range, 0.9 – 155.6/Mb). 32 samples from 24 pts had > 11/Mb and fell above the 95^th percentile for mutational burden. 10/24 pts had biallelic loss of mismatch repair (MMR) genes MSH2/6, and a further 5 pts without confirmed MMR defects had enrichment of trinucleotide signatures associated with MMR and/or were MSI high by mSINGS. The remaining 9 pts had either BRCA2 mutations or Kataegis (localized hypermutation). Clinical data was available for 10/15 MMR defective pts. Median age was 73.6 y. At diagnosis, 70% had Gleason score ≥8, 50% with M1 disease, median PSA was 22.8 (6.8 – 820). PSA50 with ADT (n = 8) or ADT + docetaxel (n = 2) was 100% in the castration sensitive setting. 5 pts had ENZ, 4 ABI + P, and 1 cabazitaxel in 1L CRPCT. Comparing the MMR defective with the control cohort, median time from ADT to CRPC was 9.1 m (95% CI 6.9 – 11.4) vs. 18.2 m (95% CI 15.1 – 21.3), p = 0.001; 1L CRPCT duration was 3.9 m (95% CI 1.3 – 6.5) vs. 8.4 m (95% CI 7.2 – 9.6), p = < 0.001; median OS was 13.1 m (95% CI 0.33 – 25.9) vs. 40.1 m (95% CI 32.4 – 47.8), p < 0.001. Conclusions: HM and MMR defects can be identified in a liquid biopsy. Although these pts can have poor outcomes with standard therapy, ctDNA may help selection for immunotherapy.