A set of methodology and process of variable network topology optimization at the middle and micro view level for large-scale special events congestion prevention was proposed based on the basic model, in order to reduce the risk and the combating costs of road traffic congestion. The variable topology models of roadway sections and intersections were established respectively, with the minimum composite saturation degree as the optimization objective of the roadway section variable topology, and the minimum sum of volume ratios of various phases on the approach as the optimization objective of the intersection variable topology. The concept and the identification principles of the key roadway sections of the variable network topology were proposed, the two-way traffic importance weights of the affected roadway sections were set, and the intersection variable topology was used as the joint and auxiliary strategy for roadway section variable topology combined with the traffic operation and congestion evolution characteristics of the large-scale special events. The optimization model solution methods were also studied. The empirical study of the proposed method was based on the leaving traffic of a large-scale special event held in Capital Indoor Stadium in Beijing. Results show that the implementation of variable topology optimization in the study area can prevent congestion caused by special event attracted traffic, which is conducive to solving the problem of serious imbalances of two-way traffic directions. The improvement percentage of composite saturation degree on all relevant roadway sections was more than 10%, indicating that the proposed method can effectively adjust the short-term structural mismatch of network traffic supply and demand, and optimize the transportation infrastructure stock under existing conditions to avoid the waste of resources caused by infrastructure increases, and it is a fine management oriented method.