Abstract: Small world property is one of the most important properties of complex networks. The theory of small world network is introduced into wireless sensor networks (WSNs), which is very important to optimize topology and improve the network performance of wireless sensor networks. A new approach RSLCS (Removing Superfluous Links and Creating Shortcuts) is proposed to construct wireless sensor networks with small world phenomenon. This method first selectively removes some links by the analysis of clustering coefficient, then creates shortcuts based on the Sink node. Two steps alternately operate until the two characters of small world (high clustering coefficient and small average path length) achieve the optimal. Simulation results show that this method not only can improve the clustering coefficient and reduce the average path length, which make the network have obvious small world characteristics, but also simplify the topology structure of the network and make the network have good invulnerability under low redundancy conditions.