An urban green space network provides safe and green exercise routes for residents. This study selected Nanchang as the study area. Using fitness application data, we explored the mobility of people exercising in the network, i.e., the flow of green exercise. Spatial analysis based on social networks, GIS, and the gravity model was used to analyze the nodes, network characteristics, and mechanism of the flow of green exercise. The results show that there were differences in the hierarchy and attractiveness of nodes. Distance had an important influence on green exercise. It was found that walkers moved between adjacent parks, runners visited more parks and corridors, while cyclists covered longer distances and preferred to explore suburban green spaces. The length of the exercise route in green spaces had a positive effect on mobility. Because of the many combinations of patches and corridors, three flow subnetworks were formed. In addition, the green space network expanded the scope of exercise services in the central node. The management of green spaces should pay attention to the social value of urban green space networks and create a hierarchical and interconnected green space for exercisers.