Abstract: To make an efficient path-finding engine, a heuristic optimization path-finding algorithm was proposed for resolving the time and space complexity problems of a searching algorithm in a large application system. The algorithm was based on the classical Dijkstra algorithm and introduced the decision mechanism in AI into path-finding. To decrease the number of nodes to search, cost-function was incorporated into this algorithm and used to decide the path-finding policy, that was, which nodes were searched firstly. The condition of getting the optimal solution from this algorithm was put forward and proved. These two algorithms were tested comparatively.