Using location information to assist routing is often proposed as an efficient means to achieve scalability in large mobile ad hoc networks (MANET). This paper proposes an algorithm, named as Cluster-Based Location Aided Routing (CLAR), a scalable and efficient routing algorithm for MANET. CLAR runs on top of a one-hop cluster cover of the MANET, which can be created and maintained by, for instance, the Least Cluster Change (LCC) algorithm. It has been proven that LCC can maintain a cluster cover with a constant density of clusterheads with the minimal update cost. CLAR then utilizes nodes’ location information to improve the network layer performance of routing. The location information of destination node is used to predict a smaller isosceles triangle, rectangle, or circle request zone, which is selected according to the relative location of the source and the destination, that covers the estimated region where the destination may locate. Instead of searching the route in the entire network blindly, CLAR confines the route searching space into a much smaller estimated range. Simulation results have shown that CLAR outperforms other protocols significantly in route set up time, routing overhead, mean delay and packet collision, and simultaneously maintains low average end-to-end delay, high success delivery ratio, low control overhead, as well as low route discovery frequency.