Some of the most critical impacts of climate change will probably occur due to changes in the extremes of the hydrological cycle (droughts and floods), yet the expected changes in future precipitation and its extremes are currently quite uncertain. Scenarios of future change are derived from simulations with global and regional climate models, and our confidence in scenarios of future precipitation extremes rests on our assessment of the reliability of those models. We propose to evaluate these models by careful comparison of their simulations with observed variations in extreme precipitation events. Our comparisons will go far beyond simple comparisons of the statistics of simulated and observed events, but will in addition assess the linkages between atmospheric circulation variability (wind speeds and directions, stability, convergence and uplift), atmospheric humidity, and resulting precipitation extremes in both models and the real world. We will then extend the assessment to consider whether including changes in greenhouse gases (and other external climate influences) in the model simulations improves the agreement with changes detected in the observational data.