The Indian Bend Wash (IBW) flood-control project relies on a greenbelt to carry floods through Scottsdale, Arizona, USA. The greenbelt is characterized by a chain of shallow artificial lakes in a larger floodplain of irrigated turf, which has been protected from encroaching urban development. As such, this urban stream–floodplain complex can be divided into three subsystems: artificial lakes, channelized stream segments, and floodplain. We conducted experiments to evaluate which, if any, of these subsystems were important sites of denitrification, and to explore factors controlling denitrification rates. Denitrification enzyme activity (DEA) bioassays were conducted on sediments from eight lake and six stream segments as well as soil samples from eight floodplain transects. Mass-specific potential denitrification rates were significantly higher in lakes than in streams or floodplains. Nutrient limitation bioassays revealed that nitrate (NO3−) limited denitrification in lake sediments, a surprising finding given that NO3−-rich groundwater additions frequently raised lake NO3− concentration above 1 mg N/L. Experiments on intact lake cores suggested that denitrification was limited by the rate NO3− diffused into sediments, rather than its availability in overlying water. Floodplain denitrification was limited by water content, not NO3− or C, and irrigation of soils stimulated denitrification. We constructed a N budget for the IBW stream–floodplain complex based on our experimental results. We found that both lakes and floodplains removed large quantities of N, with denitrification removing 261 and 133 kg N·ha−1·yr−1 from lake sediments and floodplain soils, respectively, indicating that lakes are hotspots for denitrification. Nevertheless, because floodplain area was >4.5 times that of lakes, floodplain soils removed nearly 2.5 times as much N as lake sediments. Given the desert's low annual precipitation, a finding that floodplain soils are active sites of denitrification might seem implausible; however, irrigation is common in urban landscapes, and it elevated annual denitrification in IBW. Based on our results, we conclude that construction of artificial lakes created hotspots while application of irrigation water created hot moments for denitrification in the stream–floodplain complex, demonstrating that management decisions can improve the ability of urban streams to provide critical ecosystem services like N retention.