Plants with recycle systems have complex dynamics and can present challenging control problems. In this paper, the dynamics of recycle systems were examined by analyzing a two-stage heating process with recycle. It was found that, when a positive feedback arrangement is used to represent the linear dynamics of systems with recycle, the recycle-path time constant and recycle fraction must be carefully included. Also, the overall recycle dynamics can be represented by seasonal time-series models. The overall transfer function for the recycle system might be nonrational, thus complicating controller design. Three methods of applying model-based predictive control to recycle processes were compared. The first method was a Taylor series expansion used to linearize the process model, the second method used a seasonal time-series model to represent the process, and the third method was a recycle compensator that removed the effect of recycle dynamics. Although all three methods gave good control, practical considerations determine the method that should be used.