Abstract: The radiated seismic energy strongly depends on how the rupture propagates during the earthquake. In this study, we calculated the radiated energy and apparent stress, from the known slip distribution described by Brunersquo;s source function which is commonly used in the strong motion modeling, resulting from the composite source model. A new technique based on the far-field energy integrand over a simple finite fault is developed to estimate S-wave energy radiation with associated composite source model. Comparing with point source summation proposed by Anderson in which , where is the energy radiated by S-wave, and are the scalar seismic moment and shear rigidity, respectively, and are the static and dynamic stress drops, respectively, the new result is numerically given by , where (k=1, 2, 3,hellip;, n) are the subevent size, CS and q are constants which can be determined from seismic moment conservation used previously by Zeng et al. and Anderson. As Rivera and Kanamori pointed out that, in such case, the integration of the energy flux from any point on the fault depend not only the slip function at such point, but also on the slip function everywhere over the fault plane. Moreover, we discussed the frictional overshoot and undershoot behaviors inherited from Andersonrsquo;s consideration and our current solution for the composite source model, and the scaling relation of radiated energy to the ratio of given by Anderson may be incorrect in the source energy estimation. For comparing purpose, we developed composite source model to simulate the 1976 Mw7.6 Tangshan Earthquake and calculate the radiated S-wave energies based on theAndersonrsquo;s solution and our new solution. The results show that, forAndersonrsquo;s solution, the energy conservation occurs when , and the corresponded near-fault particle velocity or PGV (peak ground velocity) reaches about2m/s which are much higher than real observation. In contrast, the result derived from our new solution indicates that the energy conservation occurs at which is similar to the theoretical consideration of ( is called apparent stress and ), and the corresponded near fault PGV is about 0.9~1.3m/s which is much smaller than that fromAndersonrsquo;s solution. We suggest that, for the strong motion simulation, the seismic moment alone is insufficient to quantify the ground motion level, the radiated energy or apparent stress derived from earthquake source inversion should be an additional constraint condition in the future development of numerical algorithm of strong motion prediction.