Aquatic plants can dominate whole-stream metabolism and cause large diel fluctuations in dissolved oxygen (DO). In the lower Yakima River (Washington State, USA), water stargrass (Heteranthera dubia) grows across the entire stream channel. Its nighttime respiration depletes DO, such that concentrations are insufficient for migrating salmonids (Oncorhynchus spp.). Plant removal is one approach to increase DO concentrations, but this has not been tested much in larger rivers, where distinct hydrology may lead to different ecosystem functional responses than in small streams and lakes. We aimed to determine 1) if plant removal increases nighttime DO, and 2) whether mechanical or flow-driven removal is more effective. We removed ~ 20,000 m² of plants from a free-flowing and a dammed reach with a mechanical harvester. We continuously monitored DO in both reaches by deploying sensors upstream and downstream of the removal area. We then calculated the difference between daily minimum DO at the downstream and upstream sensors, both before and after removal. We also observed a flow-driven plant removal, in which we compared daily minimum DO between a high-flow/low-water stargrass year and a baseline low-flow/high-water stargrass year. In the free-flowing reach, mechanical plant removal increased minimum DO concentration by an average of 0.49 mg/L (range: -0.40–1.09 mg/L), but plant removal did not change DO in the dammed reach, where a smaller proportion of plants were removed. The flow-driven plant removal increased mid-summer daily minimum DO by an average of 0.88 mg/L (range: 0.20–1.56 mg/L). However, water stargrass regrew, and by the autumn, minimum DO declined to typical concentrations. Mechanical plant removal can increase DO, but only if sufficient plants are removed. Flow-driven plant removal can reduce plant abundance across a broader spatial scale, but the effect may be transient.