With the ever increasing importance and usage of Wireless Sensor Networks (WSN), security of the network has become a major concern. As more and more crucial applications of Wireless Sensor Networks are being found, more needs to be done to ensure the safety of the data and network. Intrusion detection is a method to find and establish appropriate defenses against any malicious attack on the network. However standard intrusion detection practices alone cannot ensure the safety of the wireless sensor networks due to various constraints on the power, memory and battery life of the sensors and on many other uncertain factors. In this paper we present an intrusion detection scheme using Neural Networks, Rough sets and Firefly Algorithm (FA). © 2013 IEEE.