Open-channel metabolism is commonly measured in streams, but these measurements may be strongly biased by small inputs of low-O2 groundwater seeping into the study reach that will appear to increase community respiration (CR), but decrease gross primary production (GPP). We developed a new equation that considers groundwater inputs when calculating metabolism. When streamwater O2 concentrations are much higher than groundwater O2 concentrations, a small amount of groundwater input can substantially bias CR. For example, a 200-m, 100-L s-1 stream reach that increases in discharge by only 5% may overestimate CR by two-fold. Unlike CR, the degree of bias for GPP is unaffected by O2 concentration in groundwater, therefore it is easily calculated as a function of groundwater flux into the study reach. We applied this new correction for a cowpasture spring stream in Wyoming that increased in discharge by 50% over our study reach. Uncorrected CR was -18.8 g O2 m-2 d-1 and GPP was 11.6 g O2 m-2 d-1 in the spring stream; corrected CR was -11.4 g O2 m-2 d-1 and GPP was 16.2 g O2 m-2 d-1. Our modeling and data show that ecosystem metabolism estimates are extremely sensitive to groundwater inputs, even inputs that are difficult to measure by using mass-balance of tracers. The largest unknown is the concentration of O2 in groundwater when correcting CR. We suggest that investigators attempt to correct for the impact of groundwater on any stream where the inflows are detectable.