The high mortality rate among humans infected with certain types of Avian Influenza (AI) and the potential of a mutation that allows human-to-human transmission is a great concern for the public health. We formulate a mathematical model for the prevalence of AI in humans resulting from avian-to-human transmission. The model is important because the higher the prevalence, the higher the risk of a mutation that allows human-to-human transmission leading in a major epidemic. We formulate and analyse separate deterministic and stochastic versions of the model. Different time scale separation techniques are applied to the models. The influence of certain controllable parameters on the system equilibrium is interpreted from numerical results. Moreover, we also investigate the fluctuation of populations due to demographic stochasticity at the early stage of the prevalence of AI.