Due to the limitation of own resource and the influence of subjective attitude, the nodes in mobile peer-to-peer networks often show selfish behavior. Detecting and stimulating selfish nodes to cooperate is becoming an important research topic recently. By allowing nodes to express their subjective forward attitude freely, the detection mechanism is implemented. Not only is the objective forwarding probability determined by its own resource, but subjective forwarding probability is also determined by selfish nodes that are considered. Therefore, choosing a routing path with the highest integrated forwarding probability can alleviate the influence of selfish nodes. To motivate the excessive selfish nodes to actively cooperate, an incentive and cooperation model based on punishment mechanism is presented. The selfish nodes are punished based on the damage degree of their selfish behavior. The monitor mechanism among nodes and strict punishment mechanism assure that the strategy proof is achieved. Simulation results show that the detection and incentive mechanism can not only discover appropriate routing in the condition of every rational and selfish node has limited resource, but also stimulate selfish nodes to actively cooperate when the degree of nodes’ selfishness is high.