In this paper, we propose an enhanced approach based on first-hand reputation with allows to detected misbehavior node in Manet. The network's security is an important challenge in this kind of networks. The main objective of the misbehaving nodes in AODV routing protocol deletes all data packets that received thus, don’t transmit to their destination. In a reputation-based system, each node overhears the activity of its neighbors (transmit and receive) in first-hand or combined with second-hand. Our approach is composed of three-phase directly monitoring, calculating reputation value and node isolation. The reputation value is enhanced by the packet dropped due to other events such as overloading of queue and node availability. The node with a negative reputation will be isolated, and an alert packet will be distributed to neighboring nodes to inform, afterwards an improved local repair is started. The simulation results show that the proposed approach can detect and isolate a malicious node, which improves the packet delivery ratio and lowest increasing in the throughput, while reducing the success rate of the misbehaving nodes.