Humans bear the brunt of disaster impacts during earthquakes, making them the central focus for post-earthquake emergency response and rapid rescue operations. Predicting the number of individuals seeking shelter after an earthquake is essential for effective planning and construction of earthquake shelters, as well as for coordinating shelter and rescue operations. The number and distribution of these individuals are directly related to the seismic damage sustained by buildings. Prediction methods for building seismic damage fall into three main categories: empirical methods based on historical earthquake damage, physical structural simulation analyses, and rapid post-earthquake damage prediction techniques. Using seismic damage indices or damage matrices derived from these predictions, along with urban building and population data, two primary methods for forecasting post-earthquake shelter-seeking populations have been established, namely quantitative and proportional. Quantitative prediction methods are better suited for estimating the total number of evacuees in a city, whereas proportional prediction methods can predict the proportion of evacuees both citywide and in specific areas at present and throughout a planning period. However, both methods currently lack the necessary spatial and temporal resolution required for resilient city development. With advancements in building seismic damage prediction technologies and deeper insights into evacuation behaviors, techniques for predicting post-earthquake shelter-seeking populations are evolving toward improved temporal-spatial accuracy, more precise categorization of affected groups, and expedited predictions during post-earthquake emergencies, thereby enhancing support for the development of resilient cities.