Abstract: The selection of ground motions has significant effects on structural response evaluation in time-history analysis. It is an important problem how to select ground motions to provide accurate, efficient, and consistent estimates of “true” mean structural responses. This paper focuses on the selection and scaling of ground motions in time-history analysis for estimates of mean structural responses by target spectral matching method. In the research of spectral matching, the normalized modal-mass participation factors are used as the weight factors to consider significant contributions of higher modes. Many detailed structural analysis examples are used to check the double parameters empirical multi-band matching method and the weighted scaling method based on least squares method with more complete theory. It is proved that the double parameters empirical multi-band matching method can satisfy the requirements of the code for seismic design of buildings. The weighted scaling method is compared with the methods developed by other local scholars, the artificial wave method and the method based on NGA-West2 database. The comparative study demonstrates that the weighted scaling method has reliable accuracy for estimates of mean structural responses and wider applicability including the base-isolated buildings. Through the comparative analysis, it is found that the significant advantage of the weighted scaling method is reducing the variability of structural responses and this method is more suitable for the nonlinear time-history analysis of long period structures. In the research of target response spectra selection, a Newmark spectrum is preliminarily used as a target spectrum, which may be a solution for long and overlong period structures.