The key variables of building HVAC energy consumption are the few decisive variables among all the variables that may influence the energy consumption of building HVAC energy consumption. The HVAC key variables are important for the two commonly used energy consumption prediction models (namely white box model and black box model). The modeling process of key-variables based energy prediction is greatly simplified without excessively sacrificing accuracy compared with traditional way. The determination of key variables is a complicated process and is easily affected by the initial boundary conditions.A general identification method of key variable is proposed in this paper. The key variable are identified separately from HVAC load related variables and system related variables. This method applies both Morris method and regression method for key variable identification.Also an automatic key variable identification tool is developed based on Python and Eppy. This tool is applicable to all kinds of buildings in different climate zones. The case study shows that the key variables identified by the method proposed in this paper is able to accurately describe the variation and feature of HVAC energy consumption with a few variables.