Nutrients have been hypothesized to influence the distribution of the savanna biome through two possible mechanisms. Low nutrient availability may restrict growth rates of trees, thereby allowing for intermittent fires to maintain low tree cover; alternatively, nutrient deficiency may even place an absolute constraint on the ability of forests to form, independent of fire. However, we have little understanding of the scales at which nutrient limitation operates, what nutrients are limiting, and the mechanisms that influence how nutrient limitation regulates savanna–forest transitions. Here, I review literature, synthesize existing data, and present a simple calculation of nutrient demand to evaluate how nutrient limitation may regulate the distribution of the savanna biome. The literature primarily supports the hypothesis that nutrients may interact dynamically with fire to restrict the transition of savanna into forest. A compilation of indirect metrics of nutrient limitation suggest that nitrogen and phosphorus are both in short supply and may limit plants. Nutrient demand calculations provided a number of insights. First, trees required high rates of nitrogen and phosphorus supply relative to empirically determined inputs. Second, nutrient demand increased as landscapes approached the transition point between savanna and forest. Third, the potential for fire‐driven nutrient losses remained high throughout transitions, which may exaggerate limitation and could be a key feedback stabilizing the savanna biome. Fourth, nutrient limitation varied between functional groups, with fast‐growing forest species having substantially greater nutrient demand and a higher susceptibility to fire‐driven nutrient losses. Finally, African savanna trees required substantially larger amounts of nutrients supplied at greater rates, although this varied across plant functional groups. In summary, the ability of nutrients to control transitions emerges at individual and landscape scales, and is regulated through different mechanisms based on spatial (differences in underlying geology), temporal (stage in biome transition) and biological (species traits and community composition) variability.