Performance-based stormwater best management practice (BMP) selection should match appropriate technologies with site water quality concerns and consider the ability to meet water quality standards. Empirically derived frequency analysis of effluent flow-weighted composite event mean concentrations (EMCs) for sites in the International BMP Database are presented for total suspended solids, and total and dissolved zinc and copper for dry detention basins, retention basins, wetland basins, media filters, grassed swales, bioretention, and permeable pavement. Performance characterization amongst BMPs using the probability plots is also compared to typical percent removal type assessment. No single device emerges as providing the best treatment across all water quality parameters considered. Detention basins provide the least water quality protection amongst all devices, and will not likely satisfy the typical goal of 75% TSS removal. Retention ponds can meet the 75% objective only when median percent removal is considered, but effluent TSS EMCs are approximately twice the level as compared to bioretention cells and grassed swales for the majority of events. Most BMPs are able to meet US EPA marine water quality discharge criteria for zinc, and many are able to meet the more stringent ANZECC threshold for 95% species protection. Copper is problematic for all BMPs with respect to water quality criteria. Resurfacing with permeable pavement could prevent up to 2/3 of the contaminant concentrations from typical urban source areas. (C) 2011 Elsevier B.V. All rights reserved.