<p>The approaches used to evaluate the seismic performance of slopes have evolved over the years from simple pseudo-static analyses to sliding block analyses to more sophisticated stress-deformation analyses. This paper provides a historical review of these approaches and describes their use in developing statistical surrogate models that predict the permanent displacement of earth slopes during earthquakes. The importance of incorporating peak ground velocity (PGV) in addition to peak ground acceleration (PGA) for rigid sliding block displacement is demonstrated. An evaluation of one-dimensional decoupled and coupled analyses of flexible sliding blocks reveals that both are viable approaches to account for slope flexibility in sliding block analyses. Predictive surrogate models for the displacement of flexible sliding blocks are described that utilize PGA and PGV. Finally, surrogate models based on displacements computed using stress-deformation analyses are described. These predictive models utilize PGV alone to characterize earthquake shaking and incorporate multiple characteristics of the slope, including the depth of the critical failure surface. Among the surrogate models available, those based on dynamic stress-deformation analyses are preferred because they are based on analyses that incorporate the fewest simplifying assumptions.</p>