We propose a recipe to predict strong ground motions from scenario earthquakes which are caused by active faults. From recent developments in waveform inversion analysis for estimating rupture processes during large earthquakes, we have understood that strong ground motion is relevant to slip heterogeneity rather than total moment on the fault plane. The source model is characterized by three kinds of parameters, which we call : outer fault parameters, inner fault parameters, and extra parameters. The outer fault parameters are parameters characterizing the entire source area such as total fault length, fault width, and seismic moment. The total fault length (L) is related to the grouping of active faults, i.e. the sum of the fault segments. The fault width (W) is related to the thickness of the seismogenic zones. The total fault area S (=LW) follows the self-similar scaling relation with the seismic moment (M0) for moderate-size crustal earthquakes and departs from the self-similar model for very large crustal earthquakes. The locations of the fault segments are estimated from the geological and geomorphological surveys of the active faults and/or the monitoring of seismic activity. The inner fault parameters are parameters characterizing fault heterogeneity inside the fault area. Asperities are defined as regions that exhibit large slip relative to the average slip on the fault area. The relationship between combined area of asperities and seismic moment M0 satisfies the self-similar scaling relation. The number of asperities is related to segmentation of active faults. The rake angles of slips on the asperities should be estimated from the geological survey and/or geodetic measurements. The extra fault parameters are related to the propagation pattern of rupture within the source area. Rupture nucleation and termination are related to the geometrical patterns of the active-fault segments. The recipe proposed here is to construct the procedure for characterizing those inner, outer, and extra parameters for scenario earthquakes. Then, we have confirmed that the scaling relations for the inner fault parameters as well as the outer fault parameters are valid for characterizing earthquake sources and calculating ground motions from recent large earthquakes, such as the 1995 Kobe (Japan) earthquake, the 1999 Kocaeli (Turkey) earthquake, and the 1999 Chi-Chi (Taiwan) earthquake. We have also examined the recipe for estimating strong ground motion during the 1948 Fukui (Japan) earthquake. The simulated ground motions clearly explain the damage distribution in the Fukui basin.