Urban stormwater runoff was considered to be more contaminated than the runoff from other land uses due to vehicle and human activities. Specifically, road runoff was commonly regarded as the primary pollutant (e.g. particulates and heavy metals) in the urban environment. Effectiveness of stormwater management strategies and treatment facilities in treating road runoff were highly based on the accuracy of the stormwater quantity and quality estimation. Although direct sampling of runoff during a storm event effectively quantifies the contaminant contributions in a specific area, it is expensive and time consuming. In order to efficiently design these treatment facilities based from the typical pollutant and hydrologic processes, the operation of modelling techniques were recommended. In this research, the physical characteristics of an existing tree box filter treating road runoff were adjusted using stormwater management model (SWMM) to determine the most suitable design configuration for better volume and pollutant reduction. The catchment area (CA), storage volume (SV) and surface area (SA) of the facility was increased or decreased by an increment of 25%. These changes were simulated by SWMM and would predict the influence of larger or smaller CA, SV or SA with respect to the facility performance. The hydrologic, hydraulic and water quality data used in the modelling were gathered from the 10 storm events monitored from July 2012 to July 2013. The total suspended solids were considered as the main target pollutants for this research however, the results will be correlated with the common heavy metal constituents present in the sampled runoff. Various design installations (e.g. series and parallel installation of tree box filter) were also modelled to accurately justify the obtained design configurations. Among the different configurations simulated, increasing the original values of the CA, SV and SA between 0.75% and twice the original value would result to better performance for the system. As for the installations being modelled, a centralized system would provide better volume and pollutant reduction than a divided system.