Secure routing is the process of selecting safe and unthreatened paths in a network along which to send network traffic. There is a strong need for routing security in wireless networks, because there is a dramatic increase in routing attacks in recent years and directly affects the performance of the network. In this paper we present a Transactions Based Secure Routing (TBSR) protocol for data transmission. Here Cognitive Agents (CAs) with Behaviors-Beliefs (BB) model are placed on routers to generate beliefs over the routers functionalities w.r.t. genuinity and various transactions are classified based on the Transaction Subtlety Levels (TSL). These two factors are used in determining secure paths for data transmission of varied security levels and there by enhancing the network performance. The results are presented to demonstrate the routing delay, latency, Quality of Service(QoS), and throughput. Copyright © 2014 John Wiley & Sons, Ltd.