The invention discloses a system key variable extracting method considering wind power cluster power fluctuation. The method is characterized in that the sensitivity analysis, PV analysis and dynamic evaluation are combined to determine the key variable for reflecting the main dynamic feature of the system under the wind power cluster power fluctuation, so as to create a system key variable set in a multi-dimensional space in order to express the main features of the system under the wind power fluctuation; when determining the key variable, the overall influence of the wind power cluster and the actual control characteristic of a wind power station are fully considered, and the bus voltage and branch power which are highly influenced by the wind power cluster power fluctuation and an asynchronous generator power angle of the system losing stability under the fault in the wind power cluster sending line are extracted as the system key variables for expressing the main features of the system under the wind power cluster power fluctuation.