Strong ground motion evaluations have been based more and more on a finite-fault source rather than a point source. For inland crustal earthquakes with a long recurrence period, active fault surveys provide important information to develop an appropriate finite-fault source model (e.g. “Recipe” by Irikura and Miyake, 2015). After the 1995 Kobe earthquake, the headquarters for earthquake research promotion (HERP) has evaluated source faults from prioritized 110 major active faults among ca. 400 seismogenic active faults in Japan. As a result, the source fault length was almost equal to the active fault length. The active fault information is critical because the information is the most fundamental to construct source fault so far. The HERP has improved the method to incorporate subsurface structural datasets in order to evaluate the fault dimensions of the recent Japanese earthquakes. Based on integrated subsurface information, this paper presents the evaluation of source fault lengths of Japanese inland crustal earthquakes to examine improved HERP evaluation method. The fault length was estimated by visually superposing subsurface structural datasets on the active fault map around earthquake source regions using the GIS. The fault length of major active faults is largely consistent with the source fault length derived from the waveform inversion. While, the fault length of minor active faults is underestimated. The evaluated minor fault length from subsurface structural dataset is longer than the fault length derived from active faults. The result of this study is that the integration of subsurface dataset and active faults is effective to reduce a variability of the relation between evaluated fault length and inverted source fault length.