In this study, we conducted a series of microtremor surveys to understand the contribution of soil amplifications to the heavy structural damage of wooden houses in downtown Mashiki, Kumamoto, Japan, during the 2016 Kumamoto earthquake. We analyzed the microtremor horizontal‐to‐vertical spectral ratios (MHVRs) of each observation site. A few previous studies have demonstrated the applicability of the earthquake horizontal‐to‐vertical spectral ratios (EHVRs) to identify velocity structures. Therefore, we transformed the MHVRs into pseudo‐EHVRs (pEHVRs) using the EHVR‐to‐MHVR ratio (EMR) method. We identified the velocity structures in Mashiki, according to the diffuse field concept (DFC) for earthquake, using the pEHVRs. We also estimated the seismological bedrock motions during the mainshock based on the DFC. We then performed the seismic ground response analyses of subsurface structures, using a 1D linear analysis and an equivalent linear analysis (ELA). Finally, we obtained the distribution maps of peak ground acceleration (PGA) and peak ground velocity (PGV) for Mashiki town. We obtained the following results: (a) the thickness of the soft sediment under the southwestern part of Mashiki is deeper than that under the northeastern part; (b) the thickness of the soft sediments was a primary cause of the heavy damage to buildings of Mashiki; (c) the ground motions estimated by the ELA method seemed to be close to the observed seismic ground motions in Mashiki; (d) the distribution of the estimated PGV in Mashiki had a close relationship with the damage ratio distribution of buildings; (e) the EMR method, along with the DFC for earthquake and the 1D ELA method, successfully simulated the strong motions that occurred during the mainshock in Mashiki.