Abstract:MANET is a self-organized, infrastructure less network that is established to provide wireless communication in improvised environment. Due to the dynamic network topology, the network is more vulnerable to attacks, predominantly the routing attacks. These routing attacks are detected and the risk level is estimated, for which the necessary response actions are implemented by the Intrusion Response System. Of these response actions, the existing binary isolation and naïve fuzzy deteriorate the network performance when implemented. Hence, in this paper, An Adaptive Risk Aware Response System with Dempster Shafer theory that includes a notion of importance factors (ARSDSIF) is analysed and implemented. In this system, the importance factors are being assigned to evidences to attain the attack frequency and node reputation value of nodes in the network. These parameters (attack frequency and node reputation value) determine the risk threshold to instigate reliable time-wise isolation actions in the network. The ARSDSIF improves the overall network performance when compared to basic intrusion response actions. The effectiveness of our approach is analysed with respect to several performance metrics using NS-2 simulator.