Tailwater ecosystems are often highly productive river reaches where food webs rely disproportionately on local/autochthonous energy production. While in situ continuous dissolved oxygen (DO) data are increasingly being used to estimate gross primary productivity (GPP) and ecosystem respiration (ER), this approach is complicated in tailwaters, because the dissolved oxygen of releases is frequently far from equilibrium and there can be substantial diel variation in flow to meet hydropower or other needs. To address these idiosyncrasies, we developed a new two-station open channel metabolism model to accurately estimate reach-scale GPP and ER in tailwaters. Our approach accounts for sub-daily flow variation and considerably simplifies two-station model implementation compared to previous efforts. We apply our model to a six year DO time series and use Bayesian inverse modeling to estimate daily GPP, ER and gas exchange velocity (k600) for a 12-km reach of the Colorado River downstream of Glen Canyon Dam. We compare our model’s performance to both a more physically accurate but computationally intensive Eulerian dynamic flow model as well as a one-station model (whose assumptions are violated). These comparisons confirm that our estimates (and key derived traits) do not substantially deviate from output using the Eulerian dynamic flow model, however both two-station approaches deviate substantially from the one-station approach. We expect that our new model will help resolve an analytical bottleneck in tailwater ecology and will spur further research into the primary producers that support these ecologically, culturally and recreationally valued ecosystems.