Background: The optimal management of the primary tumour in pts with M1 at diagnosis PCa is not established. We aimed to evaluate the impact on OS of LRT (surgery or radiotherapy to the primary tumour) in de novo metastatic disease. Methods: PCa pts with M1 disease at diagnosis treated at the Royal Marsden between June 2003 and December 2011 were evaluated. LRT+ patients were defined as those that had received surgery or radiotherapy for the primary. Covariates analysed included age, diagnostic Gleason score, lines of CRPC treatment, PSA, burden of bone metastases ( ≥ 4 vs < 4 bone metastases) and ECOG PS. Kaplan-Meier analyses generated OS data. The association between LRT and OS was evaluated in univariate (UV) and multivariate (MV) Cox regression models. Results: Overall 234 pts with M1 at diagnosis were identified; 27 (11.53%) received LRT (25 XRT; 2 prostatectomy). Median time interval between diagnosis and LRT was 782 days (range 0-4130). Patients receiving LRT were younger (49 vs 61 yrs, p = 0.042), had lower baseline PSA values (68 vs 148; p < 0.001), and were more likely to have lymph node only disease (26% vs 10%; p = 0.029) and a lower burden of bone metastases with < 4 metastases (85% vs 34%;p < 0.001). Patients receiving LRT had a significantly longer survival (74.2 vs 55.1 months; HR 0.39; p < 0.001) in UV and MV cox-regression analysis (table). LRT+ remained highly prognostic, independently of disease volume at diagnosis and baseline PSA. Conclusions: LRT was associated with increased survival in patients with de novo metastatic disease, and in these analyses the prognostic utility of this LRT prognostic biomarker was independent of volume of metastatic disease at baseline and I'd baseline PSA. Other possible confounder factors may need to be taken into account when interpreting these results which require prospective validation from clinical trials such as STAMPEDE .