The quantity and timing uncertainties in demand and returns make it difficult to manage the inventory of manufacturing-remanufacturing systems. In order to keep the inventory level between reasonable intervals as much as possible, under the assumption that the inventory level process is expressed as a Lévy process with no negative jumps, the expected total discounted cost model is established by utilizing renewal process and martingale theorems, subsequently the cross-entropy method is applied to obtain the optimal production rates and adjusting threshold. Numerical results are provided to illustrate the effect of returns, demand and system parameters on the optimal control policy and the expected discounted cost.