Public bike-sharing systems provide an alternative transportation method and have been rapidly implemented throughout the world in recent years. Using these systems, individuals can find a bike station near their location, rent a bike for free or at a low cost and then ride to another station near their destination. This paper studies how to determine the number of bikes to be deployed at each station during bike repositioning in bike-sharing systems. The goal is to maximize the time interval between repositioning events, which occur when the demand for renting or returning a bike is not satisfied, or to maximize the satisfaction of demands within a fixed time interval between two repositioning events. To achieve these goals, two mathematical programming models are developed, and a numerical example is used to demonstrate the efficiency of the proposed models.