The existing network spread model ignores the information attenuation in the process of information spread, and the traditional influence maximization algorithm cannot effectively use the community structure to improve the influence spread range. To solve these problems, an algorithm of Influence Maximization on Internal Decay (IMID) based on community structure was proposed. Firstly, the community structure of a whole social network was divided and the influence range of nodes in the community was evaluated. Then, with spread probability of association points between the communities considered, the attenuation degree of information spread between nodes was calculated. Experimental and analysis results show that the proposed algorithm not only reduces the time complexity, but also obtains the influence transmission range near that of greedy algorithm, with influence coverage over 90%. Therefore, with several nodes selected as the initial nodes between the core seed node set and connected communities, information will be widely spread in the network at the minimum cost.