Stormwater runoff from existing impervious surfaces needs to be managed to protect downstream waterbodies from hydrologic and water quality impacts associated with development. As urban expansion continues at a rapid pace, increasing impervious cover, and climate change yields more frequent extreme precipitation events, increasing the need for improved stormwater management. Although green infrastructure such as bioretention has been implemented in urban areas for stormwater quality improvements and volume reductions, these systems are seldom monitored to validate their performance. Herein, we evaluate flow attenuation, stormwater quality performance, and nutrient cycling from eight roadside bioretention cells in their third and fourth years of implementation in Burlington, Vermont. Bioretention cells received varying treatments: (1) vegetation with high-diversity (7 species) and low-diversity plant mixes (2 species); (2) proprietary SorbtiveMediaTM (SM) containing iron and aluminum oxide granules to enhance sorption capacity for phosphorus; and (3) enhanced rainfall and runoff (RR) to certain cells (including one with SM treatment) at three levels (15%, 20%, 60% more than their control counterparts), mimicking anticipated precipitation increases from climate change. Bioretention water quality parameters monitored include total suspended solids (TSS), nitrate/nitrite-nitrogen (NOx), ortho-phosphorus (Ortho-P), total nitrogen (TN) and total phosphorus (TP), which were compared among bioretention cells' inflows and outflows across 121 storms. Simultaneous measurements of flow rates and volumes allowed for evaluation of the cells' hydraulic performances and estimation of pollutant load and event mean concentration (EMC) removal. We also monitored soil CO₂ and N₂O fluxes, as they represent a potential nutrient loss pathway from the bioretention cells. We determined C and N stocks in the soil media and vegetation, which are critical design elements of any bioretention, to determine the overall C and N balances in these systems. Significant average reductions in effluent stormwater volumes and peak flows were reported, with 31% of the storms events completely captured. Influent TSS loads and EMCs were well retained by all cells irrespective of treatments, storm characteristics, or seasonality. Nutrient removal was treatment-dependent, where the SM treatments consistently removed P loads and EMCs, and sometimes N as well. The vegetation and RR treatments mostly exported nutrients to the effluent. We attribute observed nutrient exports to the presence of excess compost in the soil filter media. Rainfall depth and peak inflow rate undermined bioretention performance, likely by increasing pollutant mobilization through the filter media. While the bioretention cells were a source of CO₂, they varied between being a sink and source of N₂O. CO₂ fluxes were orders of magnitude higher than N₂O fluxes. However, soil C and N, and plant C and N in biomass was seen to largely offset respiratory CO₂-C and biochemical N₂O-N losses from bioretention soil. The use of compost in bioretention soil media should be reduced or eliminated. If necessary, compost with low P content and high C: N ratio should be considered to minimize nutrients losses via leaching or gas fluxes. In order to understand trade-offs stemming from compost amendments, we conducted a laboratory pot study utilizing switchgrass and various organic soil amendments (e.g., different compost types and coir fiber) to a sandy loam soil contaminated with heavy metals and studied potential nutrient leaching and pollutant uptake. Addition of organic amendments significantly reduced metal bioavailability, and improved switchgrass growth and metal uptake potential. While no differences in soil or plant metal uptake were observed among the amendments, significant differences in nutrient leaching were observed.