In this research paper we propose a neighbor co-ordination based hybrid approach. It addresses detection of fault nodes in wireless sensor networks (WSN). In this model we consider the average of mean and median of the collected data from the neighboring sensors. The cluster head of every region collects the captured data from every neighboring node. In every fixed interval of time it computes the mean and median of the captured data and, takes its average. A rigorous experiment is carried on for testing of the proposed technique where the cluster head is assumed as the reliable node, which provides the correct data. The accuracy is measured with due respect to the data of cluster head. This model is compared with other existing models with the parameters such as; detection-accuracy (DA), false-alarm ratio (FAR) and false-positive rate (FPR). The simulated results show that the performance of proposed approach is better than the conventional methods.