Urban green space (UGS) provision across cities is often assessed from per capita quantities. However these aggregate measures say little about the actual use of UGS because they ignore the relative location of UGS and citizens. Spatial accessibility approaches consider this relative location but mostly assume that benefits happen within close proximity of residences. We challenge this assumption for three European cities comparatively, based on similarly acquired survey data. We study which factors influence how far people travel to their most used UGS, as defined by users themselves. We find that travelled distances (1.4–1.9 km) and inter-city differences are surprisingly high compared to the few hundred meters set in policy targets and accessibility analyses. We identify socio-demographic effects and a role for perceived rather than objective quality of local UGS. More than a spatial interaction trade-off between proximity and size, our results suggest that UGS visits are part of a more complex set of activities, further away from residences and with a diversity of sizes and proximities. Our results call for a re-evaluation of UGS analytical practices and provision policies beyond aggregate and accessibility perspectives, towards multi-scalar and spatially varying measures. •Distances travelled to the most used green space far exceed accessibility recommendations.•Age, occupational status, education and nationality have key effects on distances travelled.•Urban green space size and their provision in proximity seldom impact the distance travelled.•Perception of local green space provision matters more than objective measures of the provision.•Large variations across our 3 cities call for further comparative research.