Abstract: The acquirement of the global ARGO data provides unprecedented ocean data for researches on the climate prediction, a tropical dynamic ocean-statistical atmospheric coupled model is set up. In the coupled model, the atmospheric part is a statistical atmospheric model constructed according to the correlation relationship between observed windstress and sea surface temperature in the tropical Pacific. The ocean part is Zebiak-Cane oceanic model in which the subsurface temperature parameterization scheme is improved based on the ARGO data. The long-term hindcasts are done using the coupled model. In order to understand the performance of the ARGO data in the improvement of the hindcasts, the results of the coupled model are compared using the unimproved and improved ocean models respectively. It is found that the application of the ARGO data in the oceanic model greatly raises the capability of the coupled model in hindcasting the sea surface temperature (SST) in the tropical Pacific. The hindcast of 6 months in advance gives the abnormal variations of the SST in the Niño3.4 region in good agreement with the observation. The El Niño and La Niña events in the hindcasting period are successfully hindcasted. The correlation of SSTs between the hindcasts and the observations is increased substantially in the whole tropical Pacific Ocean.The improvement of the short-term climate prediction of summer rainfall in China is also investigated by using the global atmosphere-ocean coupled model of the China National Climate Center (NCC) using the ARGO data in the Global Ocean Data Assimilation System of NCC (NCC-GODAS). The results of seasonal hindcasts for the summer precipitation in China for the case of the summer of 2002 and also for the summers from 1998 to 2003 are showed respectively. It is found that when the ARGO data is included in the NCC-GODAS, the distribution of the hindcasted summer precipitation in China is closer to the observation. The area of the positive correlation between the hindcasted and the observed summer precipitations is enlarged. It demonstrates that hindcasts for the summer precipitation in China with ARGO data in the NCC-GODAS is much better than those without ARGO data.