Residential water use constitutes a major part of urban water demand, and has be gaining importance in the urban water supply. Considering the complexity of residential water use system, an agent-based social simulation, i.e. the Residential Water Use Model (RWUM), is developed in this paper to capture the behavioral characteristics of residential water usage. By disaggregating total water demands down to constituent end-uses, this model can evaluate heterogeneous consumer responses on water, taking into account the factors of market penetration of watersaving technologies, regulatory policies, economic development, as well as social consciousness and preferences. Also, uncertainty analysis technique is innovatively applied in this agent-based model for parameter calibration and model robust testing. According to the case study in Beijing, this model can provide insights to water management agency in evaluating different water usage polices, as well as estimations for potential water saving for future infrastructure development planning.