Detached eddy simulation (DES) has been widely applied in crosswind stability simulations of trains in recent years. As DES is a hybrid Reynolds Averaged Navier–Stokes (RANS)/large eddy simulation approach, the choice of the RANS model associated with DES is a key factor for an accurate numerical simulation. However, the influence of the RANS model on the flow around trains was not fully investigated in previous researches. In this study, DES with the Spalart–Allmaras (SA) model (SA-DES) and shear stress transport (SST) k−ω model (SST-DES) have been investigated owing to their ability to predict the surface pressure, aerodynamic forces, and the flow field around a 1/25th scale Class 390 train subjected to crosswinds. Numerical simulation results were validated with experimental data. Results show that both SA-DES and SST-DES predict similar trends of the mean flow field around the train. However, there were considerable differences observed in the position of separation points and consequently the separation and attachment lines on the roof and bottom of the train body. The SST-DES results correlated more closely to the experimental data than SA-DES for pressure coefficient on the roof and leeward surface at almost all loops. A slight difference in the side force and roll moment coefficients and a considerable difference in the lift force coefficient were observed for SA-DES and SST-DES. The side force coefficients calculated using SST-DES remain within the experimental uncertainty, whereas the lift force coefficients deviated greatly due to the omission of some underbody geometrical features. Compared to the experimental data, the SST-DES performs better than SA-DES. Therefore, the SST k−ω model is recommended for the RANS model associated with DES.