Background: Every oncologist caring for patients(pts) with a diagnosis of mCSPC at some point finds themselves deciding whether to use docetaxel (DOCE) in 1^st line or later, a difficult selection often based on perceived tolerability. In1^st line mCSPC, both novel hormonal therapies (NHTs) and DOCE have been shown effective, with direct comparisons lacking. We conducted the present analysis to help address this issue. Methods: All pts with mCSPC diagnosis in the VA Corporate Data Warehouse (CDW) who received either DOCE or NHT (abiraterone or enzalutamide) in the mCSPC setting and the converse after disease progression to castrate-resistant prostate cancer (mCRPC) were eligible for comparison. The cohort date range was from 7/2015 to 6/2022. We then created two arms — DOCE 1^st and DOCE later — matched 1:1 exactly for age, race, total lines of therapy received, PSA at diagnosis, Gleason score (GS), Charlson comorbidity index (CI) excluding cancer diagnosis, and rurality. Results: 580 pts met study eligibility, with two 147 patient cohorts identified for matched analysis. Median age was 69 (64-73) and 70 (65-73) for the DOCE 1^st and DOCE later cohorts, respectively. In both cohorts, 66%/31% were white/black. Starting PSA was <20/≥20 in 36%/64%. 60% had GS≥8. CI was <5/≥5 in 80%/20%. 2/≥3 total lines of therapy had been given to 32%/68%. Urban/rural distribution was 74%/26%. Median follow up of 30.2/23.7 months for DOCE 1^st/DOCE later, reflects DOCE approval in mCSPC two years before approval of NHTs. Median OS were 37.9/31.6 months for DOCE 1^st/DOCE later, [HR 0.65 (0.48-0.89), p=0.012], with HRs when comparing those from each cohort with GS≥8 or with starting PSA≥20, of 0.7 (0.47-1.05), p=0.018 for GS≥8, and 0.48 (0.32-0.73) p=0.001 for starting PSA≥20, both in favor of DOCE 1^st. Duration of abiraterone treatment was greater in DOCE later with values of 284 (160-460) and 153 (90-286) days, p<0.0001 for DOCE later and DOCE 1^st, respectively. Durations of DOCE treatment was statistically insignificant between cohorts at 135 (94-141) and 93 (51-177) days, p=0.35 for DOCE 1^st and DOCE later, respectively. Surprisingly, time to mCRPC, defined as time from diagnosis to 1^st mCRPC therapy, was longer but not statistically significant for DOCE later than for DOCE 1^st, with durations of 23.7 and 20.3 months, respectively, p=0.23. Conclusions: Our data in US Veterans suggest the strategy of using DOCE upfront in those with mCSPC followed by a NHT on progression results in superior OS compared with DOCE later in mCRPC, especially in men with PSA≥20 as a surrogate for high volume disease. Concordant with data from CHAARTED and STAMPEDE arm C where single agent ADT was the reference and ADT + DOCE emerged as the SOC for high volume or high-risk disease, the 6.3-month survival difference between strategies in this analysis argues for consideration of a prospective comparative study.