Evaluation of the simulations of the Indo-Pacific warm pool by IPCC models, however, reveals systematic biases. The size of the warm pool in the models is too small; the mean warm pool SST and the maximum SST are too high.  The trend in the warm pool size is smaller than in the observation, so is the trend in the mean warm pool SST, whether measured over the last century, the last 50 years, or the last 30 years. The trend in the tropical Indo-Pacific maximum SST in the models, on the other hand, tends to be higher than in the observation.  Possible causes for these biases are investigated. In light of recent work in understanding the time-mean effect of ENSO events on the climatological state, one possible cause that stands out is the lack of nonlinearity in the simulated ENSO.  Nino3 SST anomaly in most models has a normal distribution, in contrast with the non-normal distribution seen in the observations. The skewness of Nino3 SST anomaly in all models runs is either negative or very close to zero. This lack of nonlinearity in the simulated ENSO suggests that the time-mean effect from ENSO events on the upper ocean temperature of the tropical Pacific is weaker than in the observations, causing a warmer, smaller, and less regulated warm-pool in the models. Diagnostics to further test this suggestion are outlined. Preliminary results from an explicit calculation of the time-mean heating effect of ENSO events in NCAR CCSM3 will be presented.