Sodium has been found to be a scarce element needed and sought by mammals. To date, most geophagical studies have mainly concentrated on sodium in the soil with limited attention being given to the plant component. Mapping foliar sodium distribution is important to understand wildlife feeding patterns and distribution. In this study, we established whether remote sensing can be used to discriminate different levels of sodium concentration in grass. A GER 3700 spectrometer was used to measure spectral reflectance of grass in the field. Since savannah rangelands are characterized by mixed grass species, we first established the variation of foliar sodium concentration in different grass species and tested for possible effects of species-sodium interaction on spectral reflectance. Our results showed statistically significant differences between the mean reflectance for the low and medium sodium classes. No significant differences were observed between reflectance in the high sodium class and the lower classes. However, there was a significant interaction between sodium classes and species in influencing reflectance. We concluded that, in combination with knowledge of grass species distribution, hyperspectral remote sensing may be useful in classifying foliar sodium concentration in savannah rangelands. This may help to understand the distribution of mammals in some African savannahs where mineral nutrient availability is limiting.