We examine the impact of long-term cumulative exposure to fine particulate air pollution (PM2.5) on mortality among individuals aged 65 and older. Developing causal evidence on the long-term effects of PM2.5 exposure is challenging due to residential sorting, latent health, and measurement error in pollution exposure. We address these challenges by developing an instrumental variable analog to the Cox proportional hazards model. The IV leverages quasi-random variation in long-term PM2.5 exposure caused by the expansion of Clean Air Act regulations. We are currently estimating the model using longitudinal data on millions of senior citizens from the US Centers for Medicare and Medicaid Services.