Abstract: Revealing the influence mechanism of the built environment on the elderly’s bus ridership is crucial for solving travel-related traffic problems in an aging society. Based on multi-source data including bus smart card swiping data, GPS data, road network data, and point of interest （POI） data, this study constructs a Light Gradient Boosting Machine （LightGBM） model to explore the nonlinear influence effect of the built environment on the elderly’s departure and arrival bus passenger flow across multiple time periods. The results indicate that indicators such as the number of bus routes, the number of bus stops, and public transport supply capacity have high relative importance and positively drive bus passenger flow. POI service facilities covering commercial residence, medical health care, shopping services and other categories present high importance for both the elderly’s departure and arrival bus passenger flow, with significant threshold effects. Bus accessibility indicators have limited influence on the results. By contrast, road density shows high relative importance and obvious nonlinear effects, revealing that road network convenience is critical to increase elderly bus passenger flow. This study thoroughly analyzes the influence mechanism of the built environment on the elderly’s bus passenger flow, providing a scientific reference for the optimal design and decision-making of age-friendly conventional public transport services.