Cycling is one type of physical activities with documented health and environmental benefits. Little consensus has been reached about the impacts of urban greenness on cycling behavior because of the widely varying estimation techniques, especially at street scale. We objectively measured the urban greenness in two ways: overhead-view greenness by Normalized Di ff erence Vegetation Index (NDVI) and eye-level street greenness by Google Street View (GSV) images. Multilevel logistic regression models were used to examine the association between urban greenness and the odds of cycling (versus not cycling) for 5701 Hong Kong participants after controlling activity-in fl uencing built environment and individual-level covariates. We found the odds of cycling were positively associated with eye-level street greenness but not with overhead-view greenness across three bu ff er zones: 400 m, 800 m and 1600 m. In addition, the odds of cycling were negatively associated with po- pulation density, number of bus stops, and terrain slope, while positively associated with bike lane density. To build a cycling-friendly city, planners and designers might need to pay more attention to improve citizens' daily exposure to urban greenness, instead of traditional greenspace indices such as greenspace area or number of parks. The GSV technique is a novel and reliable method for measuring eye-level urban greenness with potential usage in further healthy city studies.