This study assesses the ability of 4 regional climate models (RCMs) from the Regional Model Inter-comparison Project (RMIP) and their regional multi-model ensemble, as well as their driving global climate model ECHAM5, to reproduce the summer extreme pre cipitation conditions in the years 1982 to 2000 and to project future (2052 to 2070) change over East Asian land areas under an A1B emission scenario. The results show that all models can adequately reproduce the spatial distribution of extreme heavy precipitation (R95P) with high spatial correlation (> 0.8). However, they do not perform well in simulating summer consecutive dry days (CDD). The ensemble average of multi-RCMs substantially improve model capability to simulate summer precipitation in both total and extreme categories when compared to each individual RCM. For individual RCMs, the spread of regional differences is assessed by bias distribution. The RegCM3 model simulates less extreme precipitation, in agreement with ECHAM5. This is probably due to the lack of an internal nudge process. Composite analysis of large-scale water vapor transport and geopotential height indicate that simulated R95P biases are associated with a deficiency in capturing the low-level field. For the projection of extreme wet and dry conditions under A1B emissions, most of the models predict an overall increase in heavy precipitation and CDD over East Asia, which will enhance the risk of drought disasters in the future