Cognitive Radio Networks (CRNs) are being studied intensively and gaining importance as spectrum is the heavily underutilized. CRN has the capability to exploit smartly the unutilized frequency spectrum. Recently, the research community started to work in the area of cognitive radio routing. In a flat topology, all nodes are of the same level and functionality, thus making it simple and efficient for smaller networks. However, when the network is large with sparse nodes, the routing information becomes more complex making cluster-based techniques really relevant to tackle such situations. In a cluster-based routing, all nodes in the network are dynamically organized into partitions called groups or clusters. In each cluster, a cluster head is chosen to help in the data transmission management and to maintain cluster membership information. This paper proposes a novel routing protocol for cognitive radio ad hoc networks (CRAHNs) based on clustering model which amends swiftly to the topological changes and establishes the routing efficiently. Our proposed approach is thoroughly evaluated through simulation study. The results state the suitability of the proposed protocol for cognitive radio ad hoc networks and demonstrate that it has better performance in terms of finding the source-destination route, reducing the amount of messages that are transmitted all over the network and minimizing the routing delay.