We leverage the timing of coalmine accidents to examine the effect of coal mining on air pollution. Safety regulations mandate that coal mining be suspended if a mine experiences an accident with 10 or more fatalities. We use a stacked difference-in-differences approach to compare counties with an accident to those experiencing an accident more than two years earlier or later. We provide evidence that the timing of accidents cannot be predicted. Next, we combine satellite-based air pollution data at the county-day level with the dates of accidents to show that on average, suspending coal mining reduces local air pollution by 8%. Changes in the level of coal consumption do not drive this reduction. We also find significant decreases in respiratory mortality after suspending coal mining with particularly large effects on vulnerable populations.