AbstractA simple and accurate remote sensing technique for the fine‐scale mapping of grassland protein densities was developed during a study of sward quality and antelope dispersion in Kenya, East Africa. Using measures of spectral reflectance and a vehicular navigation system, estimates of protein densities were recorded at rates exceeding 1000 samples per hour. Nitrogen analyses confirmed that reflectance measures were accurate predictors of protein density for a variety of grass swards but not for two species of sedge. Because the regression coefficients linking reflectance to protein density differed significantly between sward types, good estimates using this method will require a separate calibration for each type of grassland. By monitoring and correcting for variations in ambient light levels, the method can be used under a wide range of lighting conditions and for long periods. This facilitates sampling sufficiently systematically and intensively that contour plots of protein density can be constructed and then correlated with distributions of underlying abiotic factors, foraging activity of sympatric herbivores, or prior maps to characterize successional and historical chang