Adopting an agent-based approach, this paper explores the topological evolution of the Minneapolis Skyway System from a microscopic perspective. Under a decentralized decision-making mechanism, skyway segments are built by self-interested building owners. We measure the accessibility for the blocks from 1962 to 2002 using the size of office space in each block as an indicator of business opportunities. By building skyway segments, building owners desire to increase their buildings’ value of accessibility, and thus potential business revenue. The skyway network in equilibrium generated from the agent model displays similarity to the actual skyway system. The network topology is evaluated by multiple centrality measures (e.g., degree centrality, closeness centrality, and betweenness centrality) and a measure of road contiguity, roadness. Sensitivity tests such parameters as distance decay parameter and construction cost per unit length of segments are performed. Our results disclose that the accessibility- based agent model can provide unique insights for the dynamics of the skyway network growth.