AbstractHuman walking is a decisive excitation when checking vibration serviceability of several kinds of structures. Advanced models of walking load are needed to obtain realistic serviceability checks. This paper includes the study of the spatiotemporal parameters of human walking, which have a strong effect on the structural dynamic response. An advanced statistical model intended to simulate the temporal evolution of these parameters is sought. The modeling integrates previous related findings in biomechanics and results of a campaign of treadmill tests. An original model, in which the deviations of step interval are decomposed into a deterministic part due to asymmetry and a random part described by a second-order autoregressive process, is first established to model the intrasubject evolution of the step interval. The parameters of this model are formulated as a function of the walking speed, which constitutes a crucial innovation. The intersubject variability of the parameters of the previous model is then modeled statistically. The capability of the calibrated model to reproduce the experimental data is verified by comparing statistically the dynamic response of a virtual bridge in a variety of natural frequencies and walking speeds. The proposed model is more versatile than existing models because it is not restricted to a given distribution of walking speed, but it can be used in simulations with different distributions. Moreover, the application of the model can be extended to real-life walking, in which the average step speed is variable.