Motorized vehicles are one of the main anthropogenic sources of aerosols in modern cities across the world. Road green belts can purify air and improve air quality. Urban vegetation is considered to potentially filter fine and ultra-fine dust particles (particulate matter, PM₁₀ and smaller) from the air through a deposition process that takes place on leaf surfaces. Because of concern regarding pollution in urban streets, there is increasing interest in the ability of vegetative treatment systems to mitigate the consequences of pollution. Performance of these systems can depend on the characteristics of the green belt, for example, the type of vegetation, width, canopy density and porosity, among others. Generally, green road belts have been used to promote the spread of air pollutants outside of the road, while non-motorized vehicles roads have been disregarded. The focus of this study was on the efficiency of trees and hedges that are parallel to roads to reduce air pollution along a non-motorized vehicles road. We investigated the effect of the green belt structure on the pollutant concentration above roads for non-motorized vehicles with the aim of providing a basis for the design of a road green belt pattern. Concentrations of total suspended particulate (TSP) and PM₁₀ were measured along both sides of an experimental green belt in Yingze Street, Taiyuan city. Field measurements were taken using a horizontally ground-mounted tripping device of 10 m in length, which was covered with shade cloth, branches and leaves, to generate a boundary layer for the experimental vegetation canopy. Porosity, height and width were manipulated by installing different layers of shade cloth. The results clearly showed an effect of the green belt on pollutant dispersion in a street side. The most important factor affecting TSP was porosity, followed by height, and width. The most important factor affecting PM₁₀ concentrations was height, followed by porosity and width. The concentration of PM₁₀ was significantly correlated with the porosity and height of vegetation canopy. The optimum porosity, height and width were 20%, 350 cm and 250 cm, respectively. The percentage removals of TSP and PM₁₀ were 46.58% and 42.58% under the optimum green belt design. The effects of different combinations of each green belt factor on pollutant removal (TSP and PM₁₀) were 20%, 30%, 40% and 10% for porosity, 350 cm, 250 cm, 150 cm and 50 cm for height, and 250 cm, 350 cm, 150 cm and 50 cm for width. By applying an appropriate structure for urban road green belts, we can improve air quality over roads for non-motorized vehicles. For intra-urban arterial roads, green belts of low planting density and tree species of high crown porosity and large canopy should be used to reduce the concentrations of traffic-released pollutants in non-motorized roads.