Abstract Recently analyzed satellite-derived global precipitation datasets from 1987 to 2006 indicate an increase in global-mean precipitation of 1.1%–1.4% decade−1. This trend corresponds to a hydrological sensitivity (HS) of 7% K−1 of global warming, which is close to the Clausius–Clapeyron (CC) rate expected from the increase in saturation water vapor pressure with temperature. Analysis of two available global ocean evaporation datasets confirms this observed intensification of the atmospheric water cycle. The observed hydrological sensitivity over the past 20-yr period is higher by a factor of 5 than the average HS of 1.4% K−1 simulated in state-of-the-art coupled atmosphere–ocean climate models for the twentieth and twenty-first centuries. However, the analysis shows that the interdecadal variability in HS in the models is high—in particular in the twentieth-century runs, which are forced by both increasing greenhouse gas (GHG) and tropospheric aerosol concentrations. About 12% of the 20-yr time intervals of eight twentieth-century climate simulations from the third phase of the Coupled Model Intercomparison Project (CMIP3) have an HS magnitude greater than the CC rate of 6.5% K−1. The analysis further indicates different HS characteristics for GHG and tropospheric aerosol forcing agents. Aerosol-forced HS is a factor of 2 greater, on average, and the interdecadal variability is significantly larger, with about 23% of the 20-yr sensitivities being above the CC rate. By thermodynamically constraining global precipitation changes, it is shown that such changes are linearly related to the difference in the radiative imbalance at the top of the atmosphere (TOA) and the surface (i.e., the atmospheric radiative energy imbalance). The strength of this relationship is controlled by the modified Bowen ratio (here, global sensible heat flux change divided by latent heat flux change). Hydrological sensitivity to aerosols is greater than the sensitivity to GHG because the former have a stronger effect on the shortwave transmissivity of the atmosphere, and thus produce a larger change in the atmospheric radiative energy imbalance. It is found that the observed global precipitation increase of 13 mm yr−1 decade−1 from 1987 to 2006 would require a trend of the atmospheric radiative imbalance (difference between the TOA and the surface) of 0.7 W m−2 decade−1. The recovery from the El Chichón and Mount Pinatubo volcanic aerosol injections in 1982 and 1991, the satellite-observed reductions in cloudiness during the phase of increasing ENSO events in the 1990s, and presumably the observed reduction of anthropogenic aerosol concentrations could have caused such a radiative imbalance trend over the past 20 years. Observational evidence, however, is currently inconclusive, and it will require more detailed investigations and longer satellite time series to answer this question.