Abstract: Cognitive relay networks can mitigate channel fading and enhance the spectrum utilization effectively. However, relay selection is a problem of great importance to be solved. According to the features of first order partial derivative of the Signal-to-Noise Ratio (SNR) at destination node, this paper defines the cooperative efficiency and designs an iterative relay selection scheme. Considering the decentralized structure of cognitive relay networks, this paper introduces virtual timer into relay nodes and proposes the distributed algorithm based on the iterative relay selection scheme. In this algorithm, the distributed multiple relay selection is implemented through the virtual timer at individual relays and the exchange of pertinent information. The asymptotic complexity analysis and numerical simulation results prove that the proposed scheme achieves near-optimal performance with low computing efforts and communication overheads.