We evaluate the benefits of the use of a regional coupled model over its stand-alone atmospheric component when forced by reanalysis data in the simulation of the South American climate. We find that the coupling allows for a better simulation of important features of the atmospheric circulation and surface temperature. The simulated 2 meters air temperature is improved over most of the continent, the sea level pressure over the South Pacific Anticyclone area is better represented in the coupled simulation and the location of the ITCZ is improved during the austral winter. The precipitation, especially over the Andes, benefits less from the coupling, although a more realistic humidity transport leads to a reduction of the precipitation biases over extensive regions. The austral summer precipitation bias is reduced in areas such as eastern Colombia, northern Bolivia, eastern Brazil and central Argentina. For austral winter, the coupled model has a better performance in a large part of the Amazon region, in areas such as east of Peru, west Brazil, north Bolivia and south Argentina. Moreover, the regionally coupled model not only improves the simulation of important features of the observed atmospheric fields but also demonstrates good skills in reproducing the Humboldt upwelling system. Therefore, our study highlights the advantages of regional coupled models for the simulation of the South American climate, as the ocean-atmosphere interaction is of utmost importance for the circulation mechanisms that determine the climate of the region.