Occupant behavior has an important impact on building energy consumption, and the accuracy of an occupant behavior model directly affects the reliability of energy consumption simulation results. Ultra-low energy buildings are crucial to achieving building energy conservation and carbon dioxide reduction in China. In order to effectively promote the development of ultra-low energy buildings in Hot Summer and Cold Winter Climate Zones. where most residents adopt a “part-time, part-space” pattern of intermittent energy use behavior, and to solve the problem of poor indoor thermal environments and the high incremental cost of ultra-low energy, the study described in this paper takes Changsha as an example to carry out a multi-objective optimization study on ultra-low energy housing using a probabilistic behavioral model. On the basis of a probability model representing the residents’ actual behavior in Changsha, the optimization objective indicators, key variables and the technology benchmarks for ultra-low energy building were determined, then multi-objective optimization was carried out for a range of energy efficient technologies to obtain the Pareto optimal solutions. The results showed that the set of optimal solutions could reduce energy demand by 50.2 to 60.2% and reduce indoor thermal discomfort time by 3.52–11.09% compared with those of a reference base case, which just meets the requirements of the current design standard for energy efficient domestic buildings. An optimum solution for energy savings and indoor thermal comfort, along with economic costs, was identified, which can assist in decision-making by providing different preferences and provide useful reference for the design of ultra-low energy buildings in Hot Summer and Cold Winter Climate regions.