Riverine biogeochemical processes are understudied relative to headwaters, and reach‐scale processes in rivers reflect both the water column and sediment. Denitrification in streams is difficult to measure, and is often assumed to occur only in sediment, but the water column is potentially important in rivers. Dissolved nitrogen (N) gas flux (as dinitrogen (N2)) and open‐channel N2 exchange methods avoid many of the artificial conditions and expenses of common denitrification methods like acetylene block and 15N‐tracer techniques. We used membrane‐inlet mass spectrometry and microcosm incubations to quantify net N2 and oxygen flux from the sediment and water column of five Midwestern rivers spanning a land use gradient. Sediment and water column denitrification ranged from below detection to 1.8 mg N m−2 h−1 and from below detection to 4.9 mg N m−2 h−1, respectively. Water column activity was variable across rivers, accounting for 0–85% of combined microcosm denitrification and 39–85% of combined microcosm respiration. Finally, we estimated reach‐scale denitrification at one Midwestern river using a diel, open‐channel N2 exchange approach based on reach‐scale metabolism methods, providing an integrative estimate of riverine denitrification. Reach‐scale denitrification was 8.8 mg N m−2 h−1 (95% credible interval: 7.8–9.7 mg N m−2 h−1), higher than combined sediment and water column microcosm estimates from the same river (4.3 mg N m−2 h−1) and other estimates of reach‐scale denitrification from streams. Our denitrification estimates, which span habitats and spatial scales, suggest that rivers can remove N via denitrification at equivalent or higher rates than headwater streams. Key Points We adapted dissolved N2 methods and a novel modeling approach to estimate denitrification in rivers The water column accounted for 0–85% and 39–85% of combined denitrification and respiration Rivers can remove nitrogen via denitrification at equivalent or higher rates than headwaters