Plant carbon/nitrogen ratio (C:N) exerts significant control over net primary production (NPP) for most biomes, yet remote quantification at ecosystem scales is often hindered by coarse spatial resolution and by the influence of variable plant water content on spectral absorption. Consequently, remote sensing--based estimates for ecosystem properties can be masked by hectare-scale landscape patchiness and by drought stress. We approached the water content problem first by identifying those spectra sensitive to plant C:N but not sensitive to varying intensities of plant water stress under controlled conditions. Then we tested formulae developed at a plant scale with monocultures on mixed-grass prairie field plots several times during the growing season and derived an optimum rangeland C:N formula (RCNF). The RCNF was evaluated on pastures under experimental grazing treatments using mid-resolution, multispectral sensors. Delineation of canopy C:N within and between pastures was achieved under variable canopy moisture conditions using either Landsat 5 or ASTER spectral data. Landsat 5 canopy C:N ratios were estimated four times during the 2004 growing season with <14% error (RMSE = 3.1). Estimates tracked field measurements, with greater C:N ratios in April (between 30 and 34) and lower C:N in September (between 24 and 27). We also tested the RCNF on ASTER satellite data on experimental grazing treatments and found ASTER estimates were within 9.6% of field measurements (RSME = 1.5). Spatial and temporal variability among grazing treatments and collections times were similar to remote estimates despite variable plant moisture, indicating that rangeland C:N may be quantified using current, economical, satellite sensors within ¿3 C:N units.