This paper considers a multi-echelon inventory/distribution system formed by N-warehouses and M-retailers that manages a set of diverse products in a dynamic environment. Transshipment at both regional warehouses and retailers levels is allowed. A mixed integer linear programming model is developed, where product demand at the retailers is assumed to be not known. The problem consists of determining the optimal reorder policy by defining a new concept of robust retailer order in a two level programming approach, which minimizes the overall system cost, including ordering, holding in stock and in transit, transportation, transshipping and lost sales costs. A case study based on a real retailer distribution chain is solved.