We provide comprehensive estimates of air pollution's effect on short-run labor productivity for manufacturing firms in China from 1998 to 2007. An emerging literature estimates air pollution's effects on labor productivity but only for small groups of workers of particular occupations or sets of firms to ensure causality. To provide more comprehensive estimates necessary for policy analysis, we estimate effects for all but some small firms (90% of manufacturing output in China) and capture all channels by which pollution influences productivity. We instrument for reverse causality between pollution and output using thermal inversions.Our causal estimates imply that a one g/m 3 decrease in PM2.5 (SO2) increases labor productivity by 0.0084% (0.0572%) with an elasticity of -0.45 (-0.86). Lowering PM2.5 (SO2) by 1% nationwide through methods other than reducing manufacturing output would generate productivity increases of CNY 57.6 (110.5) thousand annually for the average firm and CNY 9.2 (17.6) billion annually or 0.06% (0.12%) of GDP across all firms. Accounting for output's contribution to PM2.5 (SO2) emissions leads to a net elasticity of -0.23 (-0.75).Using air quality of a nearby city conditional on wind blowing toward a focal city as an alternative instrument, we find a one g/m 3 decrease in PM10 increases productivity by 0.41% with an elasticity of -0.43 using a subsample of larger cities. Improving air quality generates substantial output and productivity benefits and these should be considered in evaluating environmental regulations and in evaluating their effect on firm competitiveness. JEL Codes: D62; Q51; Q53; R11