Understanding the structural characteristics, evolution, and associated mechanisms of different types of eco-industrial parks (EIPs) is important. Using complex network methods, this paper distinguishes planned-type and spontaneous-type EIPs by analyzing some cases of EIPs (especially China’s Xinzhuang EIP). The findings are as follows: (1) planned-type EIPs demonstrate more diverse sub-structures than spontaneous EIPs (more specifically, planned-type EIPs usually involve more obvious clustering phenomena and closer communication); (2) though in planned EIPs all kinds of centralities are reasonable for determining key enterprises, in spontaneous EIPs closeness and between centrality prove better measures than eigenvector centrality; (3) small-world and scale-free characteristics are important for enhancing flow transmission in both planned EIPs and spontaneous EIPs (specifically, planned EIPs show more significant small-world/scale-free effects than spontaneous EIPs, and scale-free characteristics with degree distribution coefficient 0.5–2.0 make EIPs less invulnerable); and (4) the Xinzhuang EIP is vulnerable in the network configuration under the attacks. It is important to maintain the stable stay and development of the key enterprises in an EIP. Comparing information and environmental scenarios for Xinzhuang EIP reveals that the park’s performance or efficiency is likely to be better in an information scenario. In a word, our results reveal the development mechanisms of EIPs, methods of evaluating key enterprises, ways of link reforms, structural stability of EIPs, and future scenarios predictions in terms of the practical situations.