Riding comfort on the seat is one of the important factors for vehicle comfort. To analyze riding comfort, there were some models for predicting human vibrations in the past studies. On the other hand, it is strongly affected by human body motion caused by vehicle excitation during driving especially low frequency, but it is difficult to predict human motion due to an unclear mechanism of muscle reflex. The purpose of this study is to construct virtual riding comfort testing simulation based on virtual prototyping of the seat. In this study, a virtual occupant model that predicts occupant motion on the seat against external excitation including muscle reflex for maintaining sitting posture constructed. Whole body was modeled as 15 segment biomechanical model (1D) with wobbling mass. Each joint has passive elastic torque and damping torque springs. Human body surface were modeled as rigid shape. The muscle reflex modeled as active joint torque with PID control for maintaining posture. Intrinsic, such as elastic spring forces that connect wobbling mass in abdominal cavity, and extrinsic parameters, such as active torque control gain, were determined by 2 step optimization process for fit measured data for calibrating. Whole body model and seat FE model (3D) with excitation calculated at the same time to make it possible to predict human body motion on the seat. As the results of validations, simulated results were well agreed with measurement data of subjects on the conditions of translational and rotational excitations. And, further development issues and visions were discussed.