This paper proposes a mixed-integer linear programming (MILP) model that is implemented based on a rolling horizon scheme to solve an aggregate production planning decision problem of a manufacturing company that produces snacks in Monterrey, Mexico. The demand of the company is characterized by trends and seasonality. The proposed solution is evaluated by means of computational experiments to determine the relation between demand uncertainty and flexibility of a production system. A 2k factorial experimental design and a multivariate regression were performed. Results show forecast bias and length of frozen period in the rolling horizon have a strong effect on total profit. The safety stock level was also found to be a significant factor, depending on the level of bias.