Abstract: This study evaluates the performance of 20 climate models in simulating the climatological characteristics of the Northern Hemisphere polar front jet (PFJ) during the cold season, using daily ERA5 reanalysis data and the historical simulation data from the Coupled Model Intercomparison Project Phase 6 (CMIP6). The potential sources of model bias are also examined. We find that the CMIP6 multi-model ensemble mean (MME) can reasonably reproduce the zonal distribution and strength of the PFJ. Upon examination of model performance in simulating the PFJ over different subregions, i.e., Europe, Asia, North Pacific, North America, and North Atlantic, the CMIP6 models show the largest biases in representing the climatological location and meridional width of PFJ over Europe. Compared with ERA5, the jet axis in MME shifts southward by approximately 3° in latitude, while the northern and southern boundaries shift southward by approximately 4°–7° in latitude, resulting in a widened meridional width of around 3° latitude. Seventy percent of CMIP6 models show the pattern correlation less than 0.8 with the seasonal-mean zonal wind over Europe in ERA5, whilst 75% of CMIP6 models have a spatial relative standard deviation larger than 1.0. A noticeable westerly bias over the southern part of the European PFJ is widely observed in all models. According to the magnitudes of PFJ-related westerly biases, we divide the models into two groups—the slightly southward-biased group (G1) and the strongly southward-biased group (G2). Results show that the G2 models significantly overestimate the cold air north of the European PFJ, leading to a stronger meridional temperature gradient south of the jet axis. Moreover, synoptic-scale transient eddy activity in the midlatitudes is more active in G2 than in G1. Biases in transient eddy activity may result in a stronger eddy kinetic energy transport towards the mean flow. Thus, the European PFJ in the models exhibits a southward shift and stronger westerly winds. By evaluating the simulation performance of CMIP6 models in climatic characteristics of PFJ, this study indicates that the European PFJ biases are likely related to the cold bias north of the PFJ and stronger transient eddy activity in the midlatitudes. According to our results, improving the representation of meridional temperature contrast and transient eddy activity in mid-high latitudes may help reduce the systematic biases of the PFJ over Europe in climate models.