AbstractIn this study, the relationship between ENSO and winter synoptic temperature variability (STV) over the Asian-Pacific-American region is examined in 26 CMIP5/6 model outputs. Compared to observations, most models fail to simulate the correct ENSO-STV relationship in historical simulations. To investigate the possible bias in the ENSO-STV simulations, two possible processes for the connection between ENSO and winter STV are examined in high pattern score (HPS) models and low pattern score (LPS) models, respectively. On the one hand, both HPS and LPS models can overall reproduce a reasonable relationship between STV and the mean-flow conditions supporting extratropical eddy development. On the other hand, only HPS models can well capture the relationship between ENSO and the development of extratropical eddies, while LPS models fail to simulate this feature, indicating that the bias in the simulated ENSO-STV relationship among CMIP5/6 models can be traced back to ENSO simulation. Furthermore, the bias of the ENSO simulation is characterized by an unreasonable SST pattern bias, with an excessive westward extension of warm SST anomalies over the western Pacific and weak warm SST anomalies over the equatorial central-eastern Pacific, resulting in the underestimation of the zonal SST anomaly gradient among models. Therefore, the ENSO pattern bias induces an unrealistic circulation and temperature gradient over the Asian-Pacific-American region, affecting the simulations of the ENSO-STV connection. In addition, the ENSO-STV relationship over the Asian-Pacific-American region is still robust in future projections based on HPS models, providing implications for the selection of future climate predictors.