Airborne pollen are important aeroallergens affecting human health. Local airborne pollen compositions can pose health-risks for the sensitized population, but at present little is known about fine-scale pollen composition patterns. The overall objective of this study is to determine local variations in tree pollen composition with passive samplers and to identify the surrounding landscape characteristics that drive them. In February–May 2017, during the tree pollen season, airborne tree pollen were measured by passive sampling at 2 m height above ground-level in 14 sites in the Flanders and Brussels-Capital region (Belgium). Non-metric multidimensional scaling was used to investigate environmental gradients that determine the pollen composition and amounts. Land cover types were identified across spatial scales ranging between 20 m and 5 km. The passive samplers detected the same pollen taxa during the same time windows as the validated volumetric Burkard samplers. Using passive samplers, we were able to measure local airborne pollen compositions. Corylus and Platanus pollen were associated to urban areas; Populus, Juglans and Fraxinus pollen to agricultural areas; forests and wetlands were sources of Alnus and Quercus pollen. Salix, Populus and Betula pollen were also mainly associated to wetlands. The landscape context drives the airborne tree pollen composition at a meso-scale (1−5 km) rather than at finer scale (20−500 m). Thus, land cover types (e.g. forest, bush land, agricultural lands and wetlands) surrounding urban areas may increase exposure to allergenic pollen in the urban area, potentially affecting the health of a large proportion of the population.