Atmospheric general circulation models (AGCMs) and coupled general circulation models (CGCMs) in the High-Resolution Model Intercomparison Project (HighResMIP) were evaluated on their ability to simulate tropical cyclone (TC) activity in the western North Pacific over its annual cycle. Specifically, we examined these models’ ability to simulate the south-north migration of mean TC genesis location. The results revealed that both types of models realistically captured TC numbers and the south-north migration of TC genesis locations in response to the annual cycle. However, TC number decreased less rapidly in the AGCMs than in both the CGCMs and observed data during the monsoon retreat period (after September). This bias was attributed to a cyclonic anomaly in the Philippine Sea in response to La Nin ̵̃a-like sea surface temperature (SST) differences between the AGCMs and the CGCMs. This cyclonic anomaly occurred when the northeasterly trade wind arose and was maintained through wind-evaporation-SST feedback.