This study explores the topological evolution of surface transportation networks, using empirical evidence and a simulation model validated on that data. Evolution is an iterative process of interaction, investment, and disinvestment. The temporal change of topological attributes for the network is also evaluated using measures of connectivity, density, heterogeneity, and connection patterns. The simulation model is validated using historical data from the Indiana interurban network. Statistical analyses suggest that the simulation model performs well in predicting the sequence of link abandonment in the interurban network as well as the temporal change of topological attributes. The simulation model is then applied on different idealized network structures. Typical connection patterns such as rings, webs, hub-and-spokes, and cul-de-sac emerge in the networks; the spontaneous organization of network hierarchies, the temporal change of spacing between parallel links, and the rise-and-fall of places in terms of their relative importance are also observed, providing evidence for the claim that network topology is an emergent property of network dynamics.