Aerosols may affect climate through the absorption and scattering of solar radiation and, in the case of Large dust particles, by interacting with thermal radiation(1-3). But whether atmospheric temperature responds significantly to such forcing has not been determined; feedback mechanisms could increase or decrease the effects of the aerosol forcing. Here we present an indirect measure of the tropospheric temperature response by explaining the ＇errors＇ in the NASA/Goddard model/data-assimilation system. These errors, which provide information about physical processes missing from the predictive model, have monthly mean patterns that bear a striking similarity to observed patterns of dust over the eastern tropical North Atlantic Ocean. This similarity, together with the high correlations between latitudinal location of inferred maximum atmospheric heating rates and that of the number of dusty days, suggests that dust aerosols are an important source of inaccuracies in numerical weather-prediction models in this region. For the average dust event, dust is estimated to heat the lower atmosphere (1.5-3.5 km altitude) by similar to 0.2 K per day. At about 30 dusty days per year, the presence of the dust leads to a regional heating rate of similar to 6 K per year.