High vehicular pollutants exposure to residents in near-road buildings raises special concerns in micro-scale urban science, as it causes severe health problems for those residents. This paper integrates a new parameter, i.e. personal intake fraction (IF_p), into computational fluid dynamics (CFD) simulations to investigate the flow and the resulted personal exposure of two-phase pollutants (CO and particular matters) in three-dimensional (3D) urban-like models. The impacts of street aspect ratios (building height/street width H/W = 0.5–1.5), viaduct settings, noise barriers and pollutant source locations were considered. 3D downward helical flows exist in the secondary streets perpendicular to the parallel approaching wind, which produces lateral pollutant transport across the interface between the main street and secondary streets. Therefore, the overall average IF_p ( ) of CO (∼0.23–0.59 ppm) in our current 3D urban models (H/W = 1) is much smaller than that in two-dimensional (2D) street canyon (∼3.25–5.21 ppm) models. Although narrower 2D street canyons usually present greater , our current 3D urban models do not show this monotone decreasing trend due to the complicated flow structures. of fine particles are always smaller than that of CO. Furthermore, if a single pollutant source is placed on the viaduct, become much smaller than that in cases with a single ground-level source. If the source location changes to the upstream secondary street, significantly decreases due to stronger local wind. Finally, of leeward-side cases usually exceeds that of windward-side cases by several times, but with viaduct settings, this leeward-to-windward ratio significantly decreases. •Personal Intake Fraction (IF_p) is used for exposure assessment in 3D urban models.•Impacts of aspect ratios in 3D models are more complicated than those in 2D ones.•IF_p in 3D urban models is one order smaller than that in 2D street canyon models.•IF_p of fine particles is always smaller than that of CO due to deposition effect.•The viaduct, noise barriers and source locations are key factors affecting IF_p.