Run-off from impervious surfaces has pervasive and serious consequences for urban streams, but the detrimental effects of urban stormwater can be lessened by disconnecting impervious surfaces and redirecting run-off to decentralized green infrastructure. This study used a before–after-control-impact design, in which streets served as subcatchments, to quantify hydrologic effectiveness of street-scale investments in green infrastructure, such as street-connected bioretention cells, rain gardens and rain barrels. On the two residential treatment streets, voluntary participation resulted in 32.2% and 13.5% of parcels having green infrastructure installed over a 2-year period. Storm sewer discharge was measured before and after green infrastructure implementation, and peak discharge, total run-off volume and hydrograph lags were analysed. On the street with smaller lots and lower participation, green infrastructure installation succeeded in reducing peak discharge by up to 33% and total storm run-off by up to 40%. On the street with larger lots and higher participation, there was no significant reduction in peak or total stormflows, but on this street, contemporaneous street repairs may have offset improvements. On the street with smaller lots, lag times increased following the first phase of green infrastructure construction, in which streetside bioretention cells were built with underdrains. In the second phase, lag times did not change further, because bioretention cells were built without underdrains and water was removed from the system, rather than just delayed. We conclude that voluntary green infrastructure retrofits that include treatment of street run-off can be effective for substantially reducing stormwater but that small differences in design and construction can be important for determining the level of the benefit. Copyright © 2015 John Wiley & Sons, Ltd.