In this study, the authors use parametric models to predict the track geometry-induced dynamic motion of a vehicle. The characteristics of vehicle dynamics in these models are directly identified through the spatial as opposed to the frequency domain. One of the former's merits is that we can identify the characteristics of vehicle dynamics with fewer observed signals than when we use spectral analysis, making it easier to obtain signals for identification. Another is that we can determine the parameters to represent dynamic behavior of a vehicle using statistical criteria. With these models, we can predict vertical acceleration of a vehicle and its wheel load as well as estimate track conditions by taking into account both ride comfort and operating safety.