Due to inventory fluctuating violently under demands and returns dependent supply disruptions, it is a very difficult problem to control the inventory effectively. Under the condition that the inventory level process represented as a Markov-modulated Lévy process, the expected cycle time and cost functions are developed by employing level crossing method, renewal process and martingale theories. Subsequently the functions are used to derive the long-run average cost rate model. Finally, the impact of the correlations between the demandsM(the returns) and the supply disruptions on the optimal inventory control policies is investigated, and the changes of the optimal policies under different disruption and return types are analyzed. Consequently, some new inventory-management insights are obtained.