Simultaneous evacuation is the most widely used evacuation strategy in buildings. However, there are other evacuation strategies that might lead to safer outcomes if selected appropriately. Different forms of evacuation result from applying time delays to phased evacuation or altering path planning. The best strategy for evacuation depends on the characteristics of the building and the circumstances of the particular emergency. A real-time evacuation path-planning model that identifies the fire hazard and proposes the best strategy of evacuation during the emergency can reduce risk and improve safety. In this paper, a model is proposed to find the safest strategy of evacuation based on the current state of the building and the emergency case. The model focuses on fire emergencies, as they are the dominant cause of fatalities in buildings compared to other types of natural and manmade disasters. The proposed model first defines a risk factor for each compartment based on the location of fire and then calculates the lowest risk path using Dijkstra algorithm. The path-planning runs on the geometric network graph (GNG), which is generated from the IFC file of the building. Furthermore, unexpected events during evacuation, e.g. another source of fire, can force the system to search for another strategy. Herein, a model is designed to monitor the building in real-time and in case of any unexpected event, changes the evacuation plan accordingly. The case study shows that the proposed model for real-time evacuation management can significantly enhance the safety level of evacuation compared to the conventional simultaneous evacuation process.