To test the capabilities of newly available instrumentation and to explore the dynamics of carbonyl sulfide (COS) as a proxy for the measurement of canopy-scale gross primary production (GPP), we conducted an experiment to measure the simultaneous net transfer of COS and CO2 between the atmosphere and a growing wheat canopy, senesced wheat, and the harvested field (located in the Southern Great Plains of the U.S.) using the eddy covariance technique. We found that during the growing season, there was a strong uptake of COS by the canopy (roughly between −10 and −40 pmol m−2 s−1) with a strong diel signal that mirrored net CO2 fluxes. After senescence and over the harvested field, we observed a strong source of COS to the atmosphere (up to +40 pmol m−2 s−1) that exhibited a weaker diel pattern, again similar to CO2. These results suggest that the dynamics of COS are more complicated than once thought, but that it may still be possible to independently derive canopy-scale GPP from direct COS measurements and to use them as model constraints on the atmospheric carbon cycle. To demonstrate this, we computed an average value of leaf relative uptake (LRU) (the scaling factor between GPP and ratios of the fluxes of COS and CO2 and ratios of the atmospheric concentrations of COS and CO2) that is in good agreement with laboratory results.