There are a large number of malicious attack nodes and collusion groups in P2P(peer to peer) network, and the existing trust models improve the P2P network environment to some extent, but the emphasis of the models are different, which are unable to fully solve large-scale malicious attacks and deception. Therefore, a P2P trust model based on equilibrium theory is presented. The model consists of the construction of the trust structure, the malicious node detection and trust speculation. It constructs trust network according to the equilibrium theory, uses the equilibrium theory to define nodes balance factor, detects balance malicious nodes by calculating the impact of malicious behavior on the network, and adopts trust inference algorithm to estimate trust nodes to prevent distrust network nodes being added. Experimental results show that the model is reliable, and the algorithm is efficient and robust.