OBJECTIVES: : To study the association between treatment for substance use and mortality among elderly fee-for-service Medicare enrollees diagnosed with advanced prostate cancer. METHODS: We used SEER-Medicare linked database to extract a cohort of men diagnosed with advanced prostate cancer between 2001 and 2004. We then identified those men with a diagnosis of substance use disorder. Medicare claims were used to identify non-pharmacological treatments for substance use.  We used Cox regression to determine the association between treatment for substance use and five-year all-cause mortality.  RESULTS: :  We identified  1,509 elderly men  with advanced prostate cancer and substance use disorders. Of these, less than five percent had any claim for non-pharmacologic treatment of substance use in the five-year period following prostate cancer.  Those treated for substance use vs. those not treated for substance use had comparable demographic and clinical characteristics. Results of Cox regression showed that  compared to those not treated for substance use,  patients receiving substance use treatment had lower hazard of all-cause mortality (HR 0.72 ; 95% CI 0.47, 0.97). Propensity score and instrumental variable approaches were used to minimize selection bias.   CONCLUSIONS: Among elderly fee-for-service Medicare elderly with prostate cancer and substance use disorder, non-pharmacologic treatment for substance use may reduce hazard of mortality.  However, the uptake of treatment for substance use treatment was very small. Substance use disorder continues to be an under-appreciated co-morbidity in elderly cancer patients.  Cancer related pain and impaired outcomes are common among advanced stage prostate cancer patients.  Integrated policies and coordinated  care is essential to effectively screen, refer and treat substance use, as well as to improve uptake of substance use treatment among elderly prostate cancer patients.