Arising from: Q. Fu et al. Nature 429, 55–58 (2004); see also communication from Gillett et al.; Fu et al. reply There has been considerable debate about changes in the temperature of the troposphere1 measured using the Microwave Sounding Unit (MSU) instrument2, 3 or radiosondes4, 5. Fu et al.6 linearly combine time series from two MSU channels to estimate vertically integrated 850–300-hPa temperatures and claim consistency between surface and free-troposphere warming for one MSU record. We believe that their approach overfits the data, produces trends that overestimate warming and gives overly optimistic uncertainty estimates. There still remain large differences between observed tropospheric temperature trends and those simulated by a climate model.