Abstract: In delay tolerant network (DTN), due to the frequent changes of network topology, there is no stable link between the end to end, so it is one of the key problems in DTN research to select the appropriate relay node for message forwarding and delivering the message to the destination node in a short time. Aiming at the blindness of relay node selection and the lack of reasonable control over message copies distribution in existing routing algorithms, an adaptive spray and wait routing algorithm based on comprehensive performance of node (CPN-ASW) is proposed: in spray phase, a new metric, called as node similarity, is used to measure the similarity degree of motion trajectory between nodes, and different relay node selection strategies are adopted according to whether the node similarity exceeds the given threshold value, and subsequently the number of message copies are adaptively allocated according to the relative utility value of nodes; in wait phase, an active forwarding strategy based on delivery probability is implemented, if a relay node has a lager delivery probability to the destination node, forwarding the message to this relay node. Simulation results show that compared with the Epidemic, Spray and Wait (SaW), EBR and PBSW, CPN-ASW can effectively control the network overhead while improving the delivery ratio and reducing the average delay.