Studies in tropical regions have shown that trees and grasses respond differently to fire, grazing and browsing. In African savannas, the responses to fire, grazing and browsing are different, determined by negative or positive correlations. Browsing may have other consequences than grazing because instead of increasing woody biomass it reduces it, causing increase in grass growth, leading to increase in fuel that results in more intense fires and decrease in woody biomass. Fire and herbivory are an important interactive disturbance factors affecting vegetation succession and the tree-grass dynamics in savanna environment. Several of the fire-herbivory interactions are landscape level effects, which is shown in 2 models. My hypothesis is that the tree-grass balances are determined by interactions of both grazing and fire. There have come new scientific data about fire and herbivory and the interaction effects on tree-grass dynamic and succession in the savanna. By analyzing my hypothesis through a comparison between hypotheses, Intermediate disturbance hypothesis, Janzen-Connell hypothesis and the Huston hypothesis, I propose several scenarios of the savanna tree-grass dynamics in East Africa, as a result of this comparison.