Building design codes recommend using nonlinear dynamic analysis for an appropriate assessment of the seismic response of several types of structures. To perform nonlinear dynamic analysis, the recorded ground motion time histories of past events with a certain seismic intensity level compatible with the regional seismological characteristic are usually scarce. Therefore, alternative selection and scaling methods on the previously recorded ground motion data set are commonly employed to overcome this issue. However, the type of method and also the random characteristics of earthquake ground motion data induce variability in the estimation of structural response. In this paper, the effect of code-based ground motion selection approaches is investigated in predicting the seismic demand of masonry structures. The dynamic behavior of the masonry prototype is simulated through a 3D numerical model. The material non-linearities are taken into account by adopting a proper constitute law. The data set of PEER is used as the pool for selection and scaling based on alternative codes, including American standard and Eurocode. Results showed that alternative code-based selection approaches reach various seismic demand levels. Keywords: Ground motion records, PEER data set, Scaling, Code-based selection, Masonry structures, Incremental dynamic analysis