Abstract: This paper describes the development and framework of a new version of LASG/IAP climate system model namely FGOALS-s (Flexible Global Ocean Atmosphere Land System model -Spectral version), which employs the recently improved version of IAP/LASG spectral AGCM namely SAMIL as its atmospheric component. With the motivation of developing a state of the art coupled climate system model suitable for East Asian monsoon climate simulations, the atmospheric component of FGOALS-s, viz. SAMIL, employs a horizontal resolution of R42, which is equivalent to 2.8125°（longitude）×1.66°（latitude）. The performances of FGOALS-s in simulating the climate mean states of the atmosphere, land surface, ocean, and seaice were evaluated using the output of a 200-years control run of FGOALS-s. The interannual variability of the tropical Pacific and the mode of North Atlantic Oscillation were also analyzed. The results indicate that the FGOALS-s model has been succeeded in controlling the long-term climate drift and has acceptable performances in realistically reproducing the climate mean states of the atmosphere, ocean and land surface. Of particular important, benefit from the relatively high horizontal resolution of the AGCM component, the observed southwest-northeastward extension of the main summertime rainbelt over East China is reasonably well reproduced in FGOALS-s, indicating the excellent performance of the model in East Asian summer monsoon simulations. One common problem of many coupled oceanatmosphere models is that the ENSO variability usually tends to be more regular than nature's, the FGOALS-s model, however, successfully reproduces an irregular ENSO cycle, although its amplitude is slightly weaker than the observation. In addition, the simulated interannual variability of the equatorial middle and western Pacific is stronger than that of the equatorial middle and eastern Pacific. In addition, the model also has reasonably well performance in reproducing the North Atlantic Oscillation mode, which is the dominant mode of wintertime climate over not only the North Atlantic domain but also the Northern Hemisphere. The main deficiency of the FGOALS-s model is that the simulated tropical SSTs are colder than the observation, while the middle latitudes are warmer than the observation. The deficiency in SST simulations has proved to be the results of the bias in net sea surface heat flux simulations, which is to great extent dominated by the bias of cloud amount simulations. Another typical deficiency of the FGOALS-s model is that the SSTs at high latitudes of North Atlantic are too cold, which leads to a larger sea-ice concentration than the observation in terms of annual mean state. In the mean time, the simulated warmer SSTs surrounding the Antarctic result in less sea-ice cover over there. Future improvements of the FGOALS-s model should focus on the treatment of cloud process in the AGCM, the meridional heat transport process of the OGCM, and the tuning of the air-sea freshwater coupling scheme.