This study aims, for hybrid electric vehicle, to develop a controller for reducing torque ripple caused by fuel combustion in the combustion engine and reaction torque from road surface. Road surface reaction force and load torque are not measurable directly because the corresponding sensors cannot be easily attached due to lack of space for attaching in the engine structure, cost and fault diagnosis system of the sensors. Moreover HEV system is nonlinear. Therefore control system for reducing torque ripple should be designed under the assumptions where road surface reaction force and torque load are supposed to be unknown, and HEV system is nonlinear. In this paper, a control method for reducing the torque ripple by controlling the HEV assist motor is proposed. This control method required the only measurable crank angle. The effectiveness of the control method is verified experimentally by implementing the control law in a HEV physical simulator. HEV physical simulator consists of a power train, an engine alternative motor and a slider crank. As a result, the torque ripple is reduced by 33 percent.