The water vapor and cloud feedbacks in 5 state-of-the-art atmospheric general circulation models have been quantified using the ENSO signal. These models include the NCAR CAM1,CAM2, the GFDL AM2p10, the NASA NSIIP model, and the Hadley Centre Model. The water vapor feedback remains to be the feedback that vary least among the models. All models appear to overestimate somewhat the water vapor feedback. (The overestimate of the water vapor feedback ranges from 20--50%) The feedback varies most among models is the feedback from the short-wave cloud forcing. With the exception of the GFDL AM2, all models underestimate the strength of the negative feedback from the short-wave forcing of clouds, though the degree of the underestimate varies greatly. The estimate of the feedback from the long-wave forcing of clouds also vary widely among models. The GFDL AM2 stands out as the model that has the best simulation of the water vapor and cloud feedbacks. These results highlight that simulating the water vapor and cloud feedbacks by GCMs remains a problem to be solved. The results also suggest understanding the inter-model differences in the simulation of these feedbacks could lead to considerable insight into the nature of the problem.