Air pollution was severe in many urban areas of the United States in the first half of the twentieth century, in part due to the burning of bituminous coal for heat. We estimate the effects of this bituminous coal consumption on mortality rates in the U.S. during the mid-20th century. Coal consumption varied considerably during the 20th century due to coal-labor strikes, wartime oil and gas restrictions, and the expansion of gas pipelines, among other reasons. To mitigate the influence of confounding factors, we use a triple-differences identification strategy that relies on variation in coal consumption at the state-year-season level. It exploits the fact that coal consumption for heating was highest in the winter and uses within-state changes in mortality in non-winter months as an additional control group. Our estimates suggest that reductions in the use of bituminous coal for heating between 1945 and 1960 decreased winter all-age mortality by 1.25 percent and winter infant mortality by 3.27 percent, saving 1,923 all age lives per winter month and 310 infant lives per winter month. Our estimates are likely to be a lower bound, since they primarily capture short-run relationships between coal and mortality.