Abstract: To support flood management practices, it is crucial to simulate detailed human evacuation processes and examine the tradeoff between system-level evacuation efficiency and individual-level fairness of flood risk exposure, known as the efficiency-fairness tradeoff.This study focuses on a residential district in Shenzhen City as the study site and develops an agent-based model (ABM) to simulate the detailed flood evacuation processes.Various scenario analyses are conducted to assess the collective evacuation performance of the community, considering factors such as the number of evacuees, strategies for allocating shelter capacity, evacuation sequences of multiple residential groups, and time delays between evacuation groups.This study also evaluates the tradeoff between efficiency and fairness during flood evacuation processes.The modeling results demonstrate that evacuation efficiency is primarily affected by shelter capacity allocation strategies and the population of evacuees, while evacuation fairness is largely affected by the evacuation sequence of multiple groups.For effective flood emergency management, authorities should consider the residential distribution and traffic network to allocate shelter capacity based on the actual evacuation demand.To achieve a balance between efficiency and fairness in flood evacuation, policymakers should account for varying flood risk levels at different locations, guide residents to evacuate in a staged order, and appropriately manage the time intervals between evacuation groups.