Several lines of evidence indicate that GCMs’ projections of extreme precipitation may not be credible. First, the temporal distributions of precipitation in GCMs generally do not match those documented in observational precipitation products. Second, the observational record generally shows larger trends in extreme precipitation than those simulated in GCMs when they are forced with observed changes in greenhouse gases over the historical period. Third, there is a relationship between the magnitude of GCM projections of future extreme precipitation change and GCM resolution, with coarser resolution models generally producing weaker changes. And finally, high resolution regional refinements of extreme precipitation events produce significantly different answers than the forcing GCMs, generally showing much larger signals. All of this evidence indicates that more work must be done to have confidence in GCM projections of change in extreme precipitation.