Hierarchical clustering is an effective method to save energy in immobile WSNs. In last decade, with the mobility applications of WSNs increasing rapidly, hierarchical clustering of mobile WSNs is becoming a high-effective and high-efficient method to save energy, such as LEACH-M, LEACH-ME, and MBC. We propose an improved LEACH protocol, which uses a fuzzy inference system (FIS) to select cluster heads (CHs). This proposed LEACH is called LEACH-FIS. This FIS considers the node’s residual energy, density, speed and direction. The interesting thing of LEACH-FIS is that, by referring the speed change (and direction change) degree relative to its neighbors, this FIS classifies the node’s speed (and direction) into the corresponding membership-function value. Simulation results show that the network lifetime of LEACH-FIS is 52.8%, 66.1% and 22.6% higher than LEACH-M, LEACH-ME, and MBC, respectively.