With the expansion of distributed multi-agent system applications and the increasing scale of the system, the characters of complex network have become an important factor in system performance. This paper makes an initial effort to find the effects of complex network characters on large-scale distributed multi-agent coordination to create a systemic analysis of the system performance and provide organization optimization algorithm designs. The study primarily investigated typical complex networks: random network, small-world network, grid network and scale-free network in multi-agent coordination on theoretical analysis and practical simulations. In theoretical analysis, the study has built the cooperative information transmission model based on Markov chain over different network topologies and compared their efficiencies on either random walk or intelligent routing model. In addition, the study explored the characters of complex network in three main coordination simulations: cooperative information transmission, multi-agent team coordination, and multi-agent network recovery. It is found that the characters of complex network such as small-world or scale-free attributes will bring significant differences in spite of the same coordination schema, and it is feasible to design some desired intelligent algorithms to take the advantage of those effects so that system performance can be promoted.