Abstract: Determination of a reasonable seismic intensity measure is very important for structural seismic performance evaluation. The seismic response of the underground structure is closely related to the deformation of the engineering site under earthquake excitation due to the constraint of the surrounding rock and soil. Besides, the burial depth of the underground structure also has critical effects on its seismic response. Therefore it is of great significance to investigate the variation of the optimal seismic intensity measure with burial depths of underground structures in engineering sites. In this paper, one-dimensional equivalent-linear earthquake site response analyses was performed by using 50 actual seismic records as the input motions to estimate the seismic response of homogeneous half-space sites and layered half-space sites. For the convenience of comparison among different numerical results, the engineering bedrock is assumed to be a linear elastic medium herein and the earthquake ground motions are input in the engineering bedrock at the same burial depth of 200 m from the ground surface. Based on the proficiency of the results, the optimal peak seismic intensity measures （peak ground acceleration PGA, peak ground velocity PGV, peak ground displacement PGD） varying with the burial depth of the site were investigated herein. The numerical results show that for the selected two types of sites, the optimal peak seismic intensity measure changes with the burial depth of the site. When the burial depth is small, the proficiency of PGA is the best. With the increase of the burial depth, the optimal proficiency changes from PGA to PGV. Moreover, although the critical burial depth corresponding to the transition from PGA to PGV are different for different sites, it exhibits a linear correlation with shear wave velocity of the engineering sites.