The paper proposes a dynamic analysis model to develop the road accident prediction model. The dynamic analysis model takes advantage of powerful prediction capability of GM (1,1) model based on grey system theory while at the same time utilizing the prediction power of Markov chain model from stochastic process theory. Furthermore, Taylor approximation method is employed to acquire the most accurate prediction. As the illustrative example, we predicted China's road accident death toll from 2006 to 2010 based on the statistical data recording of 1987 to 2005.