Abstract: In this study, a network routing algorithm, which has autonomous adaptability to network traffic conditions has been proposed. When a routing node has some different paths to a given destination, we can evaluate these paths in terms of their latency (delay time), which will be informed back from the destination node. By using latency evaluation of the path for selection, every node works as a distributed autonomous agent for adaptive routing. In forwarding packets, a routing node decides the next routing node according to the local rules with evaluation. There are 2 local rules for adaptive routing: rule 1 in which we select the next node with the shortest latency and rule 2 in which we select the next node at a probability inversely proportional to the latency value. A combination of these two rules results in the convergence to the optimal solution. Our network simulations show that this routing algorithm has good adaptability towards congested path avoidance.