Due to resource limitations in wireless sensor networks, prolonging the network lifetime has been of a great interest. An efficient routing technique is known as hierarchical routing based on clustering, in which finding the optimum cluster heads and number of them has been a challenge. In this paper, a two-level fuzzy logic is utilized to evaluate the qualification of sensors to become a cluster head. In the first level (Local Level), the qualified nodes are selected based on their energy and number of neighbors of them. Then, in the second level (Global Level), nodes’ overall cooperation is considered in the whole network with three fuzzy parameters. These parameters are centrality, proximity to base station and distance between cluster heads. Simulation results in five metrics show that the proposed approach consumes less energy and prolongs the network life time about 54 % compared with other algorithms.