By comparing the response of clouds and water vapor to ENSO forcing in nature with that in AMIP simulations by some leading climate models, an earlier evaluation of tropical cloud and water vapor feedbacks has revealed two common biases in the models: (1) an underestimate of the strength of the negative cloud albedo feedback and (2) an overestimate of the positive feedback from the greenhouse effect of water vapor. Extending the same analysis to the fully coupled simulations of these models as well as other IPCC coupled models, we find that these two biases persist. Relative to the earlier estimates from AMIP simulations, the overestimate of the positive feedback from water vapor is alleviated somewhat for most of the coupled simulations. Improvements in the simulation of the cloud albedo feedback are only found in the models whose AMIP runs suggest a positive or nearly positive cloud albedo feedback. The strength of the negative cloud albedo feedback in all other models is found to be substantially weaker than that estimated from the corresponding AMIP simulations. Consequently, although additional models are found to have a cloud albedo feedback in their AMIP simulations that is as strong as in the observations, all coupled simulations analyzed in this study have a weaker negative feedback from the cloud albedo and therefore a weaker negative feedback from the net surface heating than that indicated in observations. The weakening in the cloud albedo feedback is apparently linked to a reduced response of deep convection over the equatorial Pacific which is in turn linked to the excessive cold-tongue in the mean climate of these models. The results highlight that the feedbacks of water vapor and clouds--the cloud albedo feedback in particular--may depend on the mean intensity of the hydrological cycle. We have also examined whether the inter-model variations in the feedback from cloud albedo (water vapor) in the ENSO variability are correlated with the inter-model variations of the feedback from cloud albedo (water vapor) in global warming. While we find a weak positive correlation between the inter-model variations in the feedback of water vapor during ENSO and the inter-model variations in the water vapor feedback during global warming, we find no significant correlation between the inter-model variations in the cloud albedo feedback during ENSO and the inter-model variations in the cloud albedo feedback during global warming. The results suggest that the two common biases revealed in the simulated ENSO variability may not be necessarily carried over to the simulated global warming. These biases, however, highlight the continuing difficulty that models have to simulate accurately the feedbacks of water vapor and clouds on a time-scale we have observations.