This study examines the sensitivity of the coupled ocean-atmosphere climate of the Intra-Americas Seas (IAS; which includes the Gulf of Mexico and the Caribbean Sea) and parts of the western tropical and sub-tropical Atlantic Ocean to the prescribed bathymetry in three independent multi-decadal, high-resolution (15 km grid interval), regional coupled ocean-atmosphere model (RCM) integrations centered over the IAS. All of these RCM integrations with different prescribed bathymetries are forced by identical global atmospheric and oceanic reanalysis at the lateral boundaries. It is observed that the model integration with a smoother and coarser bathymetry in the region (RCM-C) results in more widespread sea surface temperature (SST) bias across the IAS. We also note that the bias displayed by the RCM-C simulation is analogous to the bias in the IAS ocean circulation of some Coupled Model Intercomparison Project version 5 (CMIP5) models. The models with an intermediate bathymetry (RCM-I) and finest bathymetry (RCM-F) rectify the bias in ocean transport through the Yucatan Channel relative to RCM-C but display mixed results with respect to SST bias. The RCM-C integration uses a bathymetry with a shallower Yucatan Channel, which tends to produce unrealistically weak flow through the Yucatan Channel and a weak Loop Current. However, the stronger heat transport through the Yucatan Channel in RCM-F results in significant warming of the northwestern tropical Atlantic Ocean and associated weakening of the surface easterly atmospheric winds relative to the other RCM integrations. Due to the weaker surface wind-induced shelf currents, the area comprised of western Gulf of Mexico and southern Caribbean Sea have a severe cold SST bias over the IAS in RCM-F relative to either RCM-I or RCM-C. RCM-I on the other hand significantly warms the western Gulf of Mexico. RCM-I displays the least SST bias over the IAS and the ocean transport through the Yucatan Channel is most comparable to the ocean reanalysis. We contend that RCM-I produces the most accurate simulation as a result of the coupled response to changes in bathymetry that optimizes the response of the modulation of the surface easterlies, which triggers sufficient ocean eddy activity in the western Gulf of Mexico and energizes the shelf currents. However, the uniform cold SST bias in the regional domain of all three simulations is also associated with a corresponding underestimation of the net heat flux into the ocean. The findings of this study clearly suggest that ameliorations of the SST bias in the IAS have their origins both in the atmospheric and oceanic components of the climate system.