The Model Predictive Control is a methodology very used in systems with slow dynamics like chemical process plant and supply chain. The MPC is usually used in supply chain management, under constraints like buffer limits and shipping capacities limits, based on approximations which make the future values of disturbance predicted, thus no recourse is available in the future. However, most real life applications are not only subject to constraints but also involve stochastic uncertainty. In this paper we suggest to use a Stochastic Model Predictive Control to optimize the combined costs of storage and shipping in a multi-level supply chain, to take into account the stochastic demand.