<p>With the increasing demand for high-damping dynamic analysis for nuclear power plant structures, a high-damping response spectrum consistent with the characteristics of RG1.60 spectra is in urgent demand. The direct application of the low-damping RG1.60 spectrum in high-damping analysis may result in unrealistic representation of the energy distribution at certain frequencies. It may also cause convergence difficulties during the artificial generation of time histories by fitting multi-damping spectra. In this paper, a procedure for constructing high-damping response spectra, rooted in RG1.60 spectra, was proposed. Ground motion records used to construct RG1.60 spectra were collected and normalized by PGA or PGD. For each control point of the RG1.60 spectrum, response spectrum amplification factors (RSAF) were calculated for damping ratios of 2%, 5%, 7%, 10%, 20%, and 30%, and hence the 84th percentile estimates of all RSAFs were computed. By connecting 84th percentile estimates of all control points, the trilinear high-damping spectrum was established for both the horizontal and vertical components of ground motion. The response spectra for damping ratios of 5%, 7% and 10% were compared with RG1.60 spectra, from which the feasibility of the construction procedure was validated. For further comparison at high damping ratios, the 5% damping RG1.60 spectrum was multiplied by three widely-used damping scaling factors (DSF), Osaki, Newmark-Hall, and Lin-Chang, respectively, to generate DSF-based response spectra. The comparison between the constructed high-damping spectra and DSF-based high-damping spectra shows differences within the frequency range between control points A and C. At the low frequency range where natural vibration frequencies of most base-isolated structures fall, the constructed high-damping spectra and the DSF-based spectra are close to each other.</p>