In this paper, the accuracy of a simplified modeling strategy for railway bridges under high-speed traffic based on a refined beam model for dynamic response prediction is investigated by comparison with a complex, fully three-dimensional transient coupled boundary element - finite element model approach. These models consider the track, structure, foundation, and subsoil, as well as the interaction between these subsystems. The fundamental modal properties of both the subsystems and the overall system are evaluated for both modeling strategies, and the results are used to refine the simplified modeling approach. The focus is on the accurate representation of the dissipative effects of wave propagation in the infinite soil domain. The predictions of the response resulting from the passage of a train from both modeling approaches are compared, highlighting the applicability and limitations of the simplified approach.