Abstract: A flocking control algorithm of multiple agents based on community division is presented to keep connectivity of multi-agent network. Firstly, the multi-agent network is divided into some communities based on hierarchical clustering. Secondly, a node importance evaluation algorithm is presented to select the most important node of each community as the information agent. Further more, a control algorithm is presented by preserving community connectivity to keep the connectivity of the whole multi-agent network in process of evolution. Finally the theoretical analysis and simulation experiments show that the proposed algorithm is effective.