Abstract: In this paper, 6 466 strong-motion records from 76 shallow crustal and 47 upper-mantle earthquakes in Japan were compiled to develop a damping correction factor （DCF） model. This study is aimed to provide a DCF that can be used to scale a design spectrum without known source and path parameters, therefore the models presented in this study did not include any terms for earthquake magnitude or source distance. Site effect on DCF was found to be significant, and a separate DCF model was constructed for the records from each of the four site classes. The total residuals were approximately separated into within- and between-event residuals, and the within-event residuals were further separated into within- and between-site components using a random effect model. Then the corresponding standard deviations were calculated to illustrate the variabilities associated with the source, path and site effects for the DCF model presented in this study. The results showed that the random errors tend to increase with spectral periods and damping ratios increasing, and the random errors associated with source effect are smaller than those associated with path and site effect at spectral periods less than 2.0 s. At most spectral periods, the random errors associated with site effect were smaller than those associated with path and other effects not explicitly modelled in this study. The between-event residuals are significantly correlated with magnitude and focal depth, while the within-site residuals were significantly correlated with source distance. Therefore, source and path parameters should be included for a model that can be used to scale a design spectrum from an earthquake with a known magnitude and source distance.