Two grassed bioretention cells were constructed in the easement of a bridge deck in Knightdale, North Carolina, USA, in October, 2009. One was intentionally undersized ('small'), while the other was full sized ('large') per current North Carolina standards. The large and small cells captured runoff from the 25- and 8-mm events, respectively. Both bioretention cells employed average fill media depths of 0.65 m and internal water storage (IWS) zones of 0.6 m. Flow-proportional, composite water quality samples were collected and analyzed for nitrogen species, phosphorus species, and TSS. During 13 months of data collection, the large cell's median effluent concentrations and loads were less than those from the small cell. The small cell's TN and TSS load reductions were 84 and 50%, respectively, of those achieved by the large cell, with both cells significantly reducing TN and TSS. TP loads were not significantly reduced by either cell, likely due to low TP concentrations in the highway runoff which may have approached irreducible levels. Outflow pollutant loads from the large and small cell were not significantly different from one another for any of the examined pollutants. The small cell's relative performance provides support for retrofitting undersized systems in urbanized areas where there is insufficient space available for conventional full-sized stormwater treatment systems.