Safety is the most important issue to rescue the lives of people and damage of assets. To achieve this goal, a fire accident existence predictor which also avoids hazard is crucial. The system is sensitive to temperature and pressure above the threshold values thereby produces different kinds of signals to stimulate that a fire may take place in the multi-store building at some time and location. In this paper, it has been developed a prototype of an autonomous wireless sensor network predicting and forecasting fire accident in a multi-store building. This prototype is a low cost, efficient and portable. It also is capable of avoiding the fire accident. The system displays when and where the event takes place and gives alarm signals in the form of visual, audible and SMS on LEDs and LCD, buzzer, and user mobile phones respectively. Approximation methods exist to provide estimates of smoke detector response based on optical density, temperature rise, and gas velocity thresholds. The objective of this study was to assess the uncertainty associated with these estimation methods.