<p>Growing evidence shows that metro systems are increasingly used to reach parks and other green spaces because they offer fast, low-cost and car-free access to amenities dispersed across the city. This widespread use has important implications for low-carbon travel and for the planning of transit and green-space networks, yet little is known about how the influence of green spaces on metro ridership varies across origin–destination (OD) flows within a transit network. To address this gap, we develop an extended Geographically Weighted Regression model, OD-GWR, that allows coefficients to vary across OD pairs rather than only at individual locations. We apply this model to 79,242 morning subway flows on a typical weekend day in Beijing, combining smart-card ridership records with census data and GIS-based green-space measures to estimate the heterogeneous effects of green-space attractiveness on metro use. The analysis reveals clusters of “popular green spaces” and “green corridors,” showing that the benefits of green spaces for metro access are unevenly distributed across the transit network. These findings highlight the need for planners to move beyond simply increasing amenities and toward understanding network-based heterogeneity of preferences and accessibility. By doing so, this study introduces OD-GWR as a novel analytical tool for spatial flow analysis and provides evidence to guide more equitable and sustainable urban-planning strategies.</p>