Excellent properties of gallium nitride (GaN) make it an ideal material for realizing gas sensors, especially for ammonia (NH3)3) detection. Although many researchers have pursued to describe the characteristics of GaN-based NH3 gas sensors by different approaches, few models have been reported. In this paper, with the consideration of the exponential distribution of interfacial states, a model for ammonia concentration detection of GaN gas sensors has been presented. The Poisson equation is applied to model the effect of defect states on the potential. By taking advantage of the current-voltage characteristics, the value of Schottky barrier height can be obtained. The concentration of the adsorbed NH3 gas is derived by exploiting the surface potential. It indicates that densities of acceptor interfacial trap states are in the order of 1011∼1012cm−2eV−1. The current increases with the NH3 concentration at the same applied voltage. In addition, detailed investigations of physical mechanisms and the analysis of the sensitivity have been depicted. It shows that the sensitivity followed an approximately exponential dependence on NH3 density. Results compared well with experimental data that verify the proposed model and simulation method.