This paper presents a novel heuristic method for adaptive routing policy which adopts three-step processes with state space reduction mechanism. Firstly, A* algorithm is introduced in the static road network to reduce those definitely useless states. Secondly, based on real-time traffic information and historic information, a heuristic incremental search method is applied to find the most optimal path at each information update period. Thirdly, those useless states from candidate paths are removed. Based on the real traffic data of Beijing, the experimental results illustrate the effectiveness and efficiency of our approach.