Considering the selective forwarding attacks behavior, based on multi-hop acknowledgment and trust evaluation(MHA-TE), a selective forwarding attacks detection method is proposed in wireless sensor networks. MHA-TE method using the multi-hop acknowledgment mechanism based on request response patterns from the source node. Through the source node send the request packets and the middle nodes reply to the response packets, the nodes which generate malicious packets loss behavior in the path are confirmed. Then using the malicious nodes as the standard of update parameters in trust evaluation mechanism, the data exchange among nodes is analyzed by the Bate distribution trust evaluation model. And the each node's trust value is obtained. Finally, by comparing the updated trust value with the corresponding trust value threshold, the determination of the malicious nodes is performed. By combining the best of both multi-hop acknowledgment and trust evaluation, these problems such as the high false alarm rate of multiply malicious nodes in the path, low adaptability of static trust value threshold and low accuracy detection rate can be solved. Compared with Two-hops, MLCM and ITEM methods, the simulation results show that the MHA-TE can detect the malicious nodes effectively with higher detection rate and lower false alarm rate, and also can reduce the network overhead to a large extent.