Does temperature affect macroeconomic performance directly? This paper attempts to bridge micro-level studies of temperature and task performance with recent macro-level studies of economic growth. We do so by presenting a model of labor supply under thermal stress and using country-level panel data to identify temperature-driven productivity impacts. We find that the effect of a hotter year on income and implied TFP varies with a region’s position relative to the optimal climate, so that a positive temperature shock leads to a drop in productivity in hot regions and a rise in cold ones. Countries with relatively low levels of air-conditioning penetration suffer larger implied TFP shocks in hotter years, suggesting that labor-related channels may be contributing considerably to the well-documented statistical relationship between temperature and income.