How El Niño-Southern Oscillation (ENSO) will change under global warming affects changes in extreme events around the world. The change of ENSO amplitude is investigated based on the historical simulations and representative concentration pathway (RCP) 8.5 experiments in phase 5 of the Coupled Model Intercomparison Project (CMIP5). The projected change in ENSO amplitude is highly uncertain with large intermodel uncertainty. By using the relative sea surface temperature (SST) as a measure of convective instability, this study finds that the spatial pattern of tropical Pacific surface warming is the major source of intermodel uncertainty in ENSO amplitude change. In models with an enhanced mean warming in the eastern equatorial Pacific, the barrier to deep convection is reduced, and the intensified rainfall anomalies of ENSO amplify the wind response and hence SST variability. In models with a reduced eastern Pacific warming, conversely, ENSO amplitude decreases. Corroborating the mean SST pattern effect, intermodel uncertainty in changes of ENSO-induced rainfall variability decreases substantially in atmospheric simulations forced by a common ocean warming pattern. Thus, reducing the uncertainty in the Pacific surface warming pattern helps improve the reliability of ENSO projections. To the extent that correcting model biases favors an El Niño-like mean warming pattern, this study suggests an increase in ENSO-related SST variance likely under global warming. [ABSTRACT FROM AUTHOR]