In the elastic dynamic response analysis of a structure, damping has an important influence and directly determines the energy dissipation and decay behavior of the structure. An accurate understanding of the applicable conditions of different damping models is of great significance for the analysis of practical engineering problems. The main characteristics of dynamic structural response analyses under subway-induced vibration excitations are: The subway vibration excitation intensity is small, and the structure usually remains elastic. Thus, the dynamic response of the structure is directly affected by the damping, and a reasonable damping energy dissipation is essential for the reliability of the analysis results; the frequency domain of subway-induced vibration is wide, and it is easy to excite the high-order vertical vibration pattern of the structure. Hence, the damping behavior of each order needs to be reasonably considered in the analysis. Different damping models are based on different basic assumptions and have different applicable conditions. Viscous damping and hysteretic damping are two types of commonly adopted damping models. In order to clarify the characteristics of different damping models in the structural dynamic response analysis under subway-induced vibration excitations, one typical representative for each type of damping models (Rayleigh damping and universal rate-dependent damping, respectively) is selected, and four common types of typical structures (concrete frame structure, steel frame structure, concrete shear wall structure, and steel frame-braced core tube structure) were analyzed. The results show that: When Rayleigh damping is used, the reference frequency range needs to be reasonably defined according to the structural dynamic characteristics and the external excitation frequency distribution. Compared with the Rayleigh damping model with the reference frequency range covering the main frequency of the external excitations, the model with the reference frequency range covering only the first 10 orders of vertical vibration mode is too narrow. In the cases of this paper, the vertical peak acceleration of the latter is 49% and 92% less than that of the former; the universal rate-dependent damping only requires direct definition of the expected damping-frequency relationship. The peak vertical roof accelerations of cases using the universal rate-dependent damping are 75%-156% of those adopting the Rayleigh damping model with a wide reference frequency range. The influence of damping models may differ for different structures and different locations of the same structure.