In this article, we compare ground‐motion prediction equations developed for the strong‐motion records from Japan by the Zhao, Zhang, et al. (2006) study (hereafter, the Zhao 2006 study), and the Zhao, Jiang, et al. (2016), Zhao, Liang, et al. (2016), and Zhao, Zhou, et al. (2016) studies (hereafter, the Zhao 2016 studies). The data quality and site conditions in the Zhao 2016 studies are much better than those in the Zhao 2006 study. The Zhao 2016 studies developed three separate models: one for shallow‐crustal and upper‐mantle (SC&UM) earthquakes, one for subduction‐interface events, and one for subduction‐slab events. We carry out statistical analyses in this study to test if the between‐event and within‐event standard deviations among the three models differ statistically. We also test the between‐site residuals among different models to justify the use of separate site terms for each model. We compare the predicted spectra from the Zhao 2006 and Zhao 2016 models. The Zhao 2006 spectra are comparable with those from the Zhao 2016 model in a number of magnitude, distance, and spectral period ranges but substantially different in some magnitude and distance ranges. The depth boundary of 25 km in the SC&UM models and in the shallow‐ and deep‐interface models leads to abrupt change in the predicted spectra at the depth boundary. The abrupt change can be considerable at some magnitude and distance ranges and can be averaged out using the depth‐scaling function recommended by Zhao, Zhou, et al. (2016).