To address the limitations of existing building energy benchmarking methods, a new method has been proposed for residential buildings. Grey relational analysis is used to define the correlation between different influencing factors(i.e., typical parameters) and total building energy consumption. The correlation is used as the weights of corresponding factors. Based on the weighted parameters, Cluster Analysis is performed to classify buildings into different groups. For each group, Accumulative Frequency Distribution is then conducted to identify its energy benchmarking value. This value also makes it possible to evaluate the energy-saving potentials of different buildings. The method was applied to a residential building energy consumption database established by the Architecture Institute of Japan. The results show that the method can classify buildings into different groups by both taking into consideration various influencing factors of building energy consumption as well as their weights. It can also determine building energy benchmarking values for each group and helps identify energy-saving potential of buildings and provide energy-saving strategies for occupants.