We study an Inventory Routing Problem in which the supplier has a limited production capacity and the
stochastic demand of the retailers is satisfied with procurement of transportation services. The aim is to
minimize the total expected cost over a planning horizon, given by the sum of the inventory cost at the
supplier, the inventory cost at the retailers, the penalty cost for stock-out at the retailers and the
transportation cost. First, we show that a policy based just on the average demand can have a total
expected cost infinitely worse than the one obtained by taking into account the overall probability
distribution of the demand in the decision process. Therefore, we introduce a stochastic dynamic
programming formulation of the problem that allows us to find an optimal policy in small size instances.
Finally, we design and implement a matheuristic approach, integrating a rollout algorithm and an
optimal solution of mixed-integer linear programming models, which is able to solve realistic size
problem instances. Computational results allow us to provide managerial insights concerning the
management of stochastic demand.