Abstract Seismic displacement of gravity walls had been studied using conventional static methods for controlled displacement design. In this study plain strain numerical analysis is performed using Plaxis dynamic program where prescribed displacement is applied at the bottom boundary of the soil to simulate the applied seismic load. Constrained absorbent side boundaries are introduced to prevent any wave reflection. The studied soil is chosen dense granular sand and modeled as elasto-plastic material according to Mohr–Column criteria while the gravity wall is assumed elastic. By comparing the resulted seismic wall displacements calculated by numerical analysis for six historical ground motions with that calculated by the pseudo-static method, it is found that numerical seismic displacements are either equal to or greater than corresponding pseudo-static values. Permissible seismic wall displacement calculated by AASHTO can be used for empirical estimation of seismic displacement. It is also found that seismic wall displacement is directly proportional with the positive angle of inclination of the back surface of the wall, soil flexibility and with the earthquake maximum ground acceleration. Seismic wall sliding is dominant and rotation is negligible for rigid walls when the ratio between the wall height and the foundation width is less than 1.4, while for greater ratios the wall becomes more flexible and rotation (rocking) increases till the ratio reaches 1.8 where overturning is susceptible to take place. Cumulative seismic wall rotation increases with dynamic time and tends to be constant at the end of earthquake.