In building energy predictions, the default occupancy is defined by the space function with fixed schedules. However, the occupancy of public buildings, such as galleries, recreational and educational buildings, have great variations during high and low seasons. In multi-functional public spaces, occupancy is related to space design features, which would determine the types of activities and occupants’ duration of presences. This research investigates the impacts of space design on occupancy in multifunctional spaces of public buildings using multiple cases. This study confirms that considering actual space design features will contribute in improving the accuracy of energy predictions in public buildings.