The static corner frequency and dynamic corner frequency in stochastic synthesis of ground motion from finite-fault modeling are introduced, and conceptual disadvantages of the two are discussed in this paper. Furthermore, the non-uniform radiation of seismic wave on the fault plane, as well as the trend of the larger rupture area, the lower corner frequency, can be described by the source spectral model developed by the authors. A new dynamic corner frequency can be developed directly from the model. The dependence of ground motion on the size of subfault can be eliminated if this source spectral model is adopted in the synthesis. Finally, the approach presented is validated from the comparison between the synthesized and observed ground motions at six rock stations during the Northridge earthquake in 1994. The static corner frequency and dynamic corner frequency in stochastic synthesis of ground motion from finite-fault modeling are introduced, and the conceptual disadvantages of the two are discussed in this paper. Furthermore, the non-uniform radiation of seismic wave on the fault plane, as well as the trend of the larger rupture area, the lower corner frequency, can be described by the source spectral model developed by the authors. A new dynamic corner frequency can be developed directly from the model. The dependence of ground motion on the size of subfault can be eliminated if this source spectral model is adopted in the synthesis. Finally, the approach presented is validated from the comparison between the synthesized and the ground and the ground motions at six rock stations during the Northridge earthquake in 1994.