[en] One of the most common hypotheses made for soil-structure interaction analyses is that the earthquake input motion is identical at all points beneath the structure. Several papers have recently shown that this assumption may be overly conservative and that the effect of wave passage is extremely important. These studies typically employ a relatively simple model, namely, the basemat is represented by a rectangular rigid foundation resting on top of the soil and connected to the soil by a continuously distributed set of soil springs. The seismic input is applied at the base of the soil springs and is assumed to be traveling at a constant wave velocity across the site. It is possible to improve on the soil/structure model by use of finite element methods; however, little is known about how to model the input seismic energy and typically a simple traveling wave is used. In this paper, we examine the available data to determine: (i) the appropriate wave velocity to use, and (ii) if the currently available analytic models are adequate. The choice of the appropriate apparent wave velocity to use in the analysis is not simple because in the near field of an earthquake, a number of complex arrivals give rise to the strong motion. These arrivals would be associated with the high apparent wave velocities of the lower layers, whereas, the surface wave arrivals would be associated with the much lower wave velocities of the near surface layers. Because strong motion instruments are triggered at some threshold level, it is not possible to make reasonable estimates of the appropriate wave velocities to use from the available earthquake data. Considerable data exists from underground nuclear explosions. This data was analyzed and shows that the appropriate velocity to use relative to wave passage as observed by various surface accelerometers is much higher than the near surface values. This high apparent wave train velocity is observed out to at least 10 times the depth of energy release