The response of linear elastic and non‐linear hysteretic systems having a single degree of freedom to recorded and simulated ground motions is studied. The objective is to evaluate whether the commonly used simulated motions are appropriate for predicting inelastic response of structures and elastic response of long period structures. Eight simulated motions were generated to model properties of horizontal ground motions recorded during four earthquakes. The simulated motions are sample functions of a stationary, Gaussian white noise process, multiplied by a temporal intensity function and passed through a linear single‐degree‐of‐freedom filter. Two versions, corresponding to parabolic and ‘standard’ base line corrections (BLC), of each of the simulated and recorded accelerograms were considered. The following general conclusions are deduced. Simulated ground motions should be subjected to the standard BLC, because it results in more reliable ground velocities and displacements, which in turn would lead to more reliable predictions of response of long period structures. Furthermore, the spectral density of the underlying random process, from which the simulated motions are obtained, should be modified to be more representative of the frequency content of recorded motions, especially in the low frequency range. Such an improved model can be expected to lead to better agreement, over a broad range of periods, in the average response spectra of simulated and recorded motions, for elastic as well as inelastic systems. Copyright © 1979 John Wiley & Sons, Ltd
