Displacement response spectrum (DRS), as the input, is of great significance to the displacement-based design just like the acceleration response spectrum to the traditional force-based design. Although the procedure of performance-based, in particular the displacement-based design has achieved considerable development, there is not a general DRS covering an enough long period range for common seismic design yet. This paper develops a systematic ground motion data processing procedure for the purpose of correcting the noise in the earthquake records and generating consistent DRS for seismic design. An adaptive algorithm is proposed to determine the cutoff frequency of the highpass digital filter. The DRS of more than 500 recorded earthquake ground motions are generated and they are classified into three groups according to the ratio of the peak ground acceleration to the peak ground velocity Γ°A=VΓ and/or the ratio of the peak ground velocity to the peak ground displacement Γ°V=DΓ: In each group, all the ground motions are normalized with respect to a selected scaling factor. Their corresponding DRS are obtained and then averaged to get the mean and standard deviation DRS, which can be used for both deterministic and probabilistic displacement-based design.