This study was motivated by the poor inventory management performance in a Brazilian food company with a high seasonal demand. It was clearly recognized that the best inventory management would depend on improvements in demand forecasting and in the production planning process itself. In order to deal with the identified problems, an aggregate production planning model based on linear programming has been developed. The model determines the monthly production rates and inventory levels of finished products as well as the work-force requirements to accomplish productions plans. A simple disaggregating method, which searches for equal run out times, translates the aggregate plans into a detailed master production schedule for a shorter horizon of three months. With the effective usage of this model, and improvements in the demand forecasting processes, a global reduction of inventory levels of both raw materials and final products can be achieved.