One of the main features of nodes in mobile ad hoc networks (MANETs) is their cooperation with neighbors to propagate data. Misusing this feature, malicious nodes cooperate with normal nodes to disrupt network operation and reduce its efficiency. These nodes attack other network nodes and prevent being detected by other nodes through using the mobility characteristic of nodes in MANETs. One of the effective ways to detect malicious nodes is using game theory. Focusing on this issue, in this paper, an approach is presented for malicious node detection using Games Theory. Data exchange between two nodes forms a game, and the history of the successful and unsuccessful sending/receiving of data, is stored to be used for future decisions. Malicious nodes are detected by the game components, sent and received data as well as information that have been stored in different stages. The experimental results indicate that if the density of malicious nodes approximately reaches 20%, and the game is repeated more than four times between neighboring nodes, the proposed approach raises the malicious node detection rate to 91%. However, if the density of malicious nodes is more than 30%, the algorithm can detect more than 87% of them after being repeated six times.