Wireless Sensor Networks (WSNs) typically consist of numerous sensor nodes deployed in an area, for the purpose of detecting the existence of 'interesting' patterns or events. Although various event detection (ED) approaches have been researched, most are limited to the specific ED scenarios. This paper proposes a generic state model with neighbourhood support (NS) from WSNs for emergency event detection, which addresses the reliability issue and takes the features of sensor data into account. The detection process involves: (a) sensor node level state model for suspicious behaviour detection; (b) NS checking to confirm or deny the event; (c) network-level fusion to characterise the event. The application of the generic model to both threshold-based event detection and tempo-spatial pattern based event detection is discussed. Simulation results demonstrate that the model can be applied in both ED categories and can improve the reliability and efficiency of the existing ED approaches.