It is known that for bike sharing systems, the flow of customers can completely change the temporal and spatial distribution of the bikes and cause an imbalance of demand and supply in the system. Thus rebalancing/redistribution of bikes is critical to ensure the efficiency of bike sharing systems. Rebalancing of bikes can be done either by users with incentive program or by operator with a fleet of rebalancing vehicles. In an operatorbased rebalancing method, the operator collects and repositions bikes in order to balance certain number of bikes to predetermined locations. The rebalancing can be static or dynamic or a combination of the static and dynamic. Static rebalancing means that the bikes are rebalanced without the interference of users' activities. Such rebalancing is usually operated during the night when no customers borrow or return bikes. In contrast, dynamic rebalancing is operated periodically in the day when the borrowing and returning of bikes continuously occur. Recently, a new type of bike sharing systems, the dockless/free- floating bike sharing system, has emerged which does not need docking stations, and therefore, it cuts a large percentage of startup investment. With the built-in GPS device, the free-floating bike sharing system allows users to leave a bike almost anywhere which, beside the flexibility, makes the rebalancing of these systems more challenging than typical station-based ones. In light of the above, a hybrid rebalancing method is developed in this project by combining user-based incentive program and operator-based rebalancing to take the advantage of both in free floating bike sharing systems. This method has been featured by a multi-objective technique to optimize the system based on two objectives, cost and service level, which helps decision makers have a better knowledge about the trade-off between these two objectives caused by their decision. In addition, capability of used tools in this method guarantees its applicability on real world scale problems. This technique has been successfully applied on the data collected from ShareABull system at USF.