Evaluating the importance of node is valuable for improving the network survivability. Due to the complexity of inter-domain policy, the existing evaluating methods which are based on static topology can not reflect the real importance of the autonomous systems (ASes) in routing. This paper is the first study to evaluate the AS importance through the best paths between the ASes from the view of dynamic routing. The more the best paths passing through an AS, the more important it is. An evaluation method based on preferred route is proposed, the complexity of which is O(l×nm) and the evaluated importance is more accurate. The time complexity is the same as the best complexity of the evaluating methods based on static topology. To verify the validity, the preferred route method is compared with two representative methods based on static topology under real routing data, which are degree method and stress centrality method respectively. The result shows that the preferred route method can discover the nodes of importance but small connections efficiently. Moreover, the evaluated importance is closer to the real AS importance than the other two methods.