Future projections of extreme ocean surface wave climates were carried out with single-model ensemble experiments of the atmospheric global climate model MRI-AGCM3.2H. The ensemble experiments of MRI-AGCM3.2H consist of four future sea surface temperature (SST) ensembles and three perturbed physics (PP) ensembles. This study showed that future changes in extreme wave heights strongly depend on the global climate model (GCM) performance to simulate tropical cyclones (TCs), indicating a need to acknowledge that results in a study that employs a low-performance model are not able to account for extreme waves associated with TCs (TC waves). The spatial distribution of future changes in non-TC extreme wave heights on the global scale was similar to that for mean wave heights; namely, wave heights increase over the middle-to-high latitudes in the Southern Ocean and central North Pacific and decrease over midlatitudes and the North Atlantic, although the magnitude of future changes for extreme wave heights is greater than for mean wave heights. The variance of future changes mainly depends on differences in physics among PP ensemble experiments rather than differences in SST ensembles. The 10-yr return wave heights of TC waves over the western North Pacific showed either an increase or a decrease of 30% for different regions, maximally. The spatial distribution of future changes in TC waves can be explained by an eastward shift of TC tracks.