An uncertainty assessment for precipitation datasets simulated by Atmosphere-Ocean Coupled General Circulation Model (AOGCM) is conducted to provide reliable climate scenario over East Asia. Most of results overestimate precipitation compared to the observational data (wet bias) in spring-fall-winter, while they underestimate precipitation (dry bias) in summer in East Asia. Higher spatial resolution model shows better performances in simulation of precipitation. To assess the uncertainty of spatiotemporal precipitation in East Asia, the cyclostationary empirical orthogonal function (CSEOF) analysis is applied. An annual cycle of precipitation obtained from the CSEOF analysis accounts for the biggest variability in its total variability. A comparison between annual cycles of observed and modeled precipitation anomalies shows distinct differences: 1) positive precipitation anomalies of the multi-model ensemble (MME) for 20 models (thereafter MME20) in summer locate toward the north compared to the observational data so that it cannot explain summer monsoon rainfalls across Korea and Japan. 2) The onset of summer monsoon in MME20 in Korean peninsula starts earlier than observed one. These differences show the uncertainty of modeled precipitation. Also the comparison provides the criteria of annual cycle and correlation between modeled and observational data which helps to select best models and generate a new MME, which is better than the MME20. The spatiotemporal deviation of precipitation is significantly associated with lower-level circulations. In particular, lower-level moisture transports from the warm pool of the western Pacific and corresponding moisture convergence significantly are strongly associated with summer rainfalls. These lower-level circulations physically consistent with precipitation give insight into description of the reason in the monsoon of East Asia why behaviors of individually modeled precipitation differ from that of observation.