The invention relates to an APT defense method based on a game system. The optimal response measure can be obtained according to the Nash equilibrium strategy, and the low efficiency caused by manual intervention on the detection of the traditional invasion system can be effectively solved. The APT defense method provided by the invention comprises the following steps: checking a fragile point of a network, and obtaining all possible attack strategies of an attacker according to a network topology structure; calculating an attacker benefit in the game system according to factors affecting the attacker benefit; calculating a defender benefit according to the factors affecting the attacker benefit; and obtaining the maximum defender benefit under the premise of Nash equilibrium.