Background Overuse of anti-cancer therapy is an important quality-of-care issue. An aggressive approach to treatment can have negative effects on quality of life and cost, but its effect on survival is not well-defined. Methods Using the Surveillance, Epidemiology, and End Results-Medicare database, we identified 7,879 Medicare-enrolled patients aged 65 or older who died after having survived at least 3 months after diagnosis of advanced non-small cell lung cancer (NSCLC) between 1991 and 1999. We used Cox proportional hazards regression analysis, propensity scores, and instrumental variable analysis (IVA) to compare survival among patients who never received chemotherapy (n = 4,345), those who received standard chemotherapy but not within two weeks prior to death (n = 3,235), and those who were still receiving chemotherapy within 14 days of death (n = 299). Geographic variation in the application of chemotherapy was used as the instrument for IVA. Results Receipt of chemotherapy was associated with a 2-month improvement in overall survival. However, based on three different statistical approaches, no additional survival benefit was evident from continuing chemotherapy within 14 days of death. Moreover, patients receiving chemotherapy near the end of life were much less likely to enter hospice (81% versus 51% with no chemotherapy and 52% with standard chemotherapy, P < 0.001), or were more likely to be admitted within only 3 days of death. Conclusions Continuing chemotherapy for advanced NSCLC until very near death is associated with a decreased likelihood of receiving hospice care but not prolonged survival. Oncologists should strive to discontinue chemotherapy as death approaches and encourage patients to enroll in hospice for better end-of-life palliative care.