Although many suggest that future diets should include more plant-based proteins, animal-sourced foods are unlikely to completely disappear from our diet. Grasslands yield a notable part of the world’s animal protein production, but thus far, there is no global insight into the relationship between current livestock stocking densities and the availability of grassland forage resources. This inhibits acting upon concerns over the negative effects of overgrazing in some areas and utilising the potential for increasing production in others. Previous research has examined the potential of sustainable grazing but lacks generic and observation-based methods needed to fully understand the opportunities and threats of grazing. Here we provide a novel framework and method to estimate global livestock carrying capacity and relative stocking density, i.e. the reported livestock distribution relative to the estimated carrying capacity. We first estimate the aboveground biomass that is available for grazers on grasslands and savannas based on the MODIS Net Primary Production (NPP) approach on a global scale. This information is then used to calculate reasonable livestock carrying capacities, using slopes, forest cover and animal forage requirements as restrictions. With this approach, we found that stocking rates exceed the forage provided by grasslands in northwestern Europe, midwestern United States, southern China and the African Sahel. In this study, we provide the highest resolution global datasets to date. Our results have implications for prospective global food system modelling as well as national agricultural and environmental policies. These maps and findings can assist with conservation efforts to reduce land degradation associated with overgrazing and help identify undergrazed areas for targeted sustainable intensification efforts.