The inherent advantages such as high robustness, low cost and high starting torque have made switched reluctance machine a strong candidate for electric vehicle applications. However, the serious vibration and acoustic noise are very troublesome. In this paper, a semi-analytical vibration prediction model is developed and an enhanced vibration reduction method is presented via random-varying turn-off angle control that is based on the mechanical property of switched reluctance machine. In this method, a random-variation-frequency sine function is adopted to make the turn-off angle vary with time. At first, a magneto-mechanical coupling model is presented, then the principle of the control strategy is introduced. Next, simulation results are presented under different operating conditions, which validate the effectiveness of the proposed method.