Background: To test whether local treatment (LT), namely radical prostatectomy (RP) or brachytherapy (BT) still confers a survival benefit versus no local treatment (NLT), when adjusted for baseline PSA (bPSA). To further examine whether the effect of LT might be modulated according to bPSA and M1 substages. Methods: Of 13906 mPCa patients within the SEER (2004-2014), 375 underwent RP, 175 BT, and 13356 NLT. Multivariable competing risks regression (MVA CRR) analyses after 1:2 propensity score matching assessed the impact of LT versus NLT on cancer specific mortality (CSM). Interaction analyses tested the association between treatment type and bPSA within different M1 substages. Results: MVA CRR analyses revealed lower CSM rates for LT (RP [HR: 0.55, CI: 0.44-0.70, P <0.001] and BT [HR: 0.63, CI: 0.49-0.83, P <0.001]) compared to NLT. A significant interaction existed between bPSA and treatment type, in M1b patients only. Here, LT conferred a survival benefit when bPSA was <60ng/mL with maximum benefit when bPSA was <40ng/mL. No survival benefit existed for M1b patients above the 60ng/mL bPSA threshold and for M1c patients, regardless of bPSA. For M1a patients, LT conferred a survival benefit compared to NLT. However, dose-response according to bPSA could not be tested, due to insufficient sample size. Conclusions: Our observations provide new insight regarding the pivotal effect of bPSA and M1 substages on CSM, when LT is contemplated. While M1a patients benefited from LT, the survival benefit was modulated by bPSA in M1b patients and no survival benefit existed in M1c patients. © 2018 Wiley Periodicals, Inc.