Abstract: In this study, the role of air–sea interactions in the simulation of the East Asian Summer Monsoon (EASM) was investigated based on reanalysis data and Model FGOALS-g3. For climatological mean, the eastward shift of the summer monsoon rain belt in AGCMs (atmospheric general circulation models) leads to dry bias over the East Asian monsoon region. While CGCMs (coupled general circulation models) present a proper position of the summer monsoon rain belt, rain in the East Asian monsoon region is still weak. Due to the absence of air–sea interaction, the fake convective responses over the Northwest Pacific region result in an eastward shift of the subtropical high. This is the main reason for the bias of AGCM in simulating the EASM precipitation. And also the deficient water vapor over land is not negligible. The two factors together contribute to over 70% of the dry error. The position bias of the subtropical high and summer monsoon rain belt is considerably reduced in CGCM, which reduces the dry bias by 36%. Less evaporation and zonal water vapor transport in CGCM simulation jointly contribute to more than 70% of dry bias. Regarding the responses to prewinter El Niño, CGCM can reproduce the Western North Pacific Anomalous Anticyclone (WNPAC) and the dipole–precipitation anomaly distribution in the El Niño decaying summer as well as the Indo-western Pacific Ocean capacitor (IPOC) which maintains the WNPAC. In the AGCM simulation, convection in the Northwest Pacific, the Bay of Bengal, and oceans surrounding the Indian Peninsula is observed to be too sensitive to sea surface temperature in the El Niño decaying summer. On the one hand, this leads to strong convective mass transport in the Northwest Pacific, inhibiting the establishment of WNPAC to some extent. On the other hand, strong convection in the North Indian Ocean can inhibit convection in the Indian Ocean, causing the inability to simulate the IPOC mode. Thus, the lack of air–sea interaction is the main reason for the inability of the AGCM to reproduce the two-season lag responses of the EASM to major El Niño.