The invention provides a voltage threshold probability-based reactive power optimizing method for a grid containing a wind power plant, which aims to solve the problem that under the condition of wind power networking, the effect of the voltage optimizing control to an automatic voltage control system is not ideal. The method comprises the steps as follows: firstly, collecting grid operation parameters, basic parameters of each wind power generator in the wind power plant, and active power output data of the wind power plant; secondly, acquiring voltage impact factor, power conversion coefficient of the wind power generators and wind power plant active power probability distribution equation through the above parameters, and further obtaining voltage threshold probability; and lastly, establishing a reactive power optimizing model on the basis of the voltage threshold probability, adopting genetic algorithm to solve the model, and obtaining a control instruction of a reactive power device. According to the invention, the voltage threshold probability-based reactive power optimizing method substitutes a conventional reactive power optimizing method for the automatic voltage control system, can enable the voltage optimizing control to the automatic voltage control system to effectively control the reactive power device in the grid under the condition of wind power networking, and ensure the safe and economic operation of the grid.