Nowadays, the network defence policy selection using game model of incomplete information ignores the type of the defender, which quantifies cost simply, resulting in unreasonable defence policies selection. Aiming at the problem, we use Bayesian game theory to model the active defence policy selection. We take the types of both the attacker and the defender into consideration. Besides, the traditional quantization method is enhanced. Then, we calculate the equilibrium of static Bayesian game. Based on the analysis of the equilibrium, we select the optimal defence policy through the prediction for attackers’ actions. The paper calculates the defence effectiveness of defence policies and provides a defence policies selection algorithm. Ultimately, we present an example to verify the effectiveness of the method and model proposed in the paper. Nowadays, the network defense policy selection using game model of incomplete information ignores the type of the defender, which quantifies cost simply, resulting in unreasonable defense policy selection. Aiming at the problem, we use Bayesian game theory to model the active defense policy selection. We take the types of both the attacker and the defender into consideration. Then, we calculate the equilibrium of static Bayesian game. Based on the analysis of the equilibrium, we select the optimal defense policy through the prediction for attackers’ actions. The paper calculates the defense effectiveness of defense policies and provides a defense policies selection algorithm. Ultimately, we present an example to verify the effectiveness of the method and model proposed in the paper.