A challenge in designing wireless sensor networks is to maximize the lifetime of the network with respect to limited resources and energy. These limitations make the network particularly vulnerable to attacks from adversaries. Denial of Service (DOS) is considered a severely damaging attack in monitoring applications when intruders attack the network and force it to lose its power and die early. There are intrusion detection approaches, but they require communications and calculations which waste the network’s limited resources. In this paper, we propose a new intrusion detection model that is suitable for defending against DOS attacks. We use the idea of energy prediction to anticipate the energy consumption of the network in order to detect intruders based on the each individual node’s excessive usage of power. Our approach does not require a lot of communications or calculations between the nodes and the cluster head. It is energy efficient and accurate in detecting intruders. Simulations show that our energy aware intrusion detection approach can effectively detect intruders based on energy consumption rate.