An online strategy-matching method for transient precaution control based on game theory is proposed，which regards the contingencies as the strategy of a virtual player and the control actions as that of a dispatcher.The payoff function is defined as the product of variation of stability margin and correction factors of strategy indicating the cost of control action and the influence of the changing environment on contingency，respectively.The method for obtaining Nash equilibrium is given by transforming the gaming model into a dual linear programming model，while the strategy with maximum expected revenue is selected as the precaution action.Finally，the method is verified effective on a New England test system.