Development of effective roadside vegetative models is important for achieving urban air-pollution-control objectives in metropolitan cities worldwide. However, there are a number of interactive factors contributing to the effectiveness of the removal of particulate matter (PM) from urban vegetation such as species composition, vegetation alignment, and environmental factors including air temperature, humidity, wind, and sources of PMs. We selected six types of roadside vegetative models including arbor, shrub, herb, arbor-shrub, arbor-herb, and arbor-shrub-herb according to three typical urban streets of Haidian District, Beijing, China to study their effectiveness on PMs. PM samplers (Dustmate) and a portable weather station (NK4500) were placed in the central part of a roadside green belt at 10 m and 20 m away from the edge of the outer lane. PM concentration and environmental factors were determined at 1.5 m and 3 m heights, respectively. The study period was from spring between mid-March and early April in 2014 when PM concentration was usually high and urban vegetation had phenologically recovered. We found that the concentration of airborne PM under a multi life form model was more stable than under a single life form vegetative model because of the effect of wind speed and relative humidity. In addition, there was a general trend showing that PMs with larger size were removed more efficiently in green belts. Vertical removal rate of airborne particulates was mainly affected by the degree of surface coverage of green belts. However, the vertical removal rate did not increase with higher surface coverage, even though it was positively correlated with air temperature. Herbs and shrubs showed stronger ability in aiding particulate matter diffusion than the other four types. Overall, the vegetative models of arbor-herb and shrubs are preferred for removing airborne PM over the other four models because of the vegetation canopy density, porosity, and species composition.