Abstract. Agent-based modeling (ABM) is a valuable tool for assessing the life-safety capacity of buildings during evacuations. This study focuses on the Kalayaan Residence Hall (KRH), a university dormitory, using Evacuation Simulation using Cognitive Agent-based modeling on Possible Earthquake (ESCAPE). ESCAPE integrates cognitive behavior, addressing the oversight in traditional ABM by considering human aspects. The 3D GIS-based model of KRH includes floors, rooms, exits, and path networks. Occupants involved in the model comprise residents, maintenance, office staff, and guards. The simulations, featuring eight scenarios, vary in agent numbers, locations, and time-based behaviors derived from survey responses. In 800 simulation runs, 79 exceeded the recommended evacuation time of 150 seconds. Analysis reveals evacuation time is influenced by the evacuee count, with over 500 occupants causing inefficiencies. The study suggests increasing occupants' knowledge about evacuation protocols and conducting regular drills to reduce panic. Recommendations include enhancing unused exits' serviceability to improve the evacuation scheme. This comprehensive approach aims to optimize KRH's evacuation efficiency by addressing cognitive aspects and refining protocols.