As a new strategy for treating excess nutrients in roadway runoff, a self-filtering roadway could be accomplished by including engineered infiltration media within a vegetated filter strip (VFS) located in the roadway shoulder. However, nutrient removal performance will depend on the design to effectively infiltrate roadway runoff and the capacity of subsurface media to sequester or remove nutrients from infiltrated runoff. The objective of this study is to test hydraulic and nutrient removal performance of a roadside VFS over varied rainfall-runoff event sizes and filter widths. Two identical 1:1 scale physical models of roadway shoulders and embankments, one containing engineered media (Treatment model) and the other without (Control model), were tested with simulated rainfall and runoff from 1- and 2-lane roadways. Overall, 32 paired hydraulic experiments and 28 paired nutrient removal experiments were completed to assess performance across frequent and extreme rainfall-runoff events. The results indicate that scalability of performance with filter width varied by parameter. Runoff generation scaled predictably with filter width, as runoff generated close to the pavement and total infiltration increased with filter length. A 6 m-wide VFS containing the engineered media infiltrated all rainfall-runoff except during the most extreme storm events (1-h storms of 76.2 mm and 50.8 mm), where respectively 35% and 22% of rainfall-runoff did not infiltrate and left the system as surface runoff. A majority of phosphorus was retained within a 1.5 m filter while nitrate removal was not observed until 6 m. The Treatment model strongly outperformed the Control model with respect to nitrate (arithmetic mean ± standard deviation of 94 ± 6% reduction vs. 23 ± 64% increase, p < .001) and total nitrogen removal (80 ± 5% vs. 38 ± 23% reduction, p < .001) due to higher rates of microbially-mediated denitrification in the Treatment model. The two models performed comparably with regard to phosphorus reduction (84 ± 9% vs. 82 ± 12% reduction). A minimum 6 m filter width is recommended to ensure sufficient infiltration of runoff and nitrogen removal. Results of this study address uncertainty regarding nutrient removal performance of VFS in urban runoff applications and highlight a potential strategy for standardizing VFS performance across varied soil properties by including engineered media within the filter.