The degree to which federal policies, such as the Clean Air Act (CAA), actually improve air quality is not fully understood. We investigate what portion of reductions in ambient fine particulate matter (PM2.5) that occurred 1999–2005 can be attributed to sulfur dioxide (SO2) and nitrogen oxide (NOx ) emissions reductions from implementation of title IV, phase 2, of the 1990 CAA Amendments. A detailed statistical model links sources and receptors over time and space to estimate the relationship between changes in emissions and observed improvements in air quality. We employ relatively transparent statistical methods incorporating uncertainty bounds to complement point estimates of the complex physico-chemical fate and transport models commonly used to estimate source-receptor relationships associated with long-range emissions transport. Monitor-specific estimates of changes in PM2.5 from changes in emissions from individual power plants are highly significant and mostly of the expected relative magnitudes for distance and direction from sources; and the model performs well on out-of-sample forecasts. Although we observe substantial model uncertainty, using our preferred specification, we estimate that the title IV, phase II emissions reduction policy implemented 1999–2005 reduced PM2.5 in the eastern USA by an average of 1.07 μg/m3, roughly 8 % (standard deviation, 0.11 μg/m3) versus a counterfactual of no change in emission rates per unit of energy input. On a population-weighted basis, the comparable reduction in PM2.5 is 0.89 μg/m3, roughly 6 %. This model presents a practical tool that can be used for policy analysis of air quality.