Due to growing awareness of non-point source pollution treatment, the performance of grass swales as a highway runoff treatment and the effect of including a grass filter strip pretreatment area adjacent to the swale were evaluated using a field-scale input/output study on a Maryland highway. Results of this comparison for 22 rainfall events over 1.5 years show significant peak reduction (50-53%), delay of the peak flow (33-34 min) and reduction of total volume (46-54%). The grass swales exhibited statistically significant removals by mean concentration of total suspended solids (41-52%), nitrite (56-66%) and zinc (30-40%), lead (3-11%), copper (6-28%) and cadmium. Other monitored nutrients (nitrate, TKN, and total phosphorus) exhibited variable removal capabilities (-1-60%), while the swales exported chloride (216-499 mg/l) at a significant level. Results suggest the pretreatment grass filter strip imparts no significant water quantity or quality improvement and that the swale itself is the most important treatment mechanism.