Several Swedish sawmill groups have recently developed product-costing systems to possibly compensate for diminishing production knowledge in recently centralized market organizations. The concept of product costs is challenging in sawmilling, since production is of a joint type – each log typically yields many products. The newly developed costing systems rely on traditional accounting-based methods and are of little use in decision-making because the resulting cost figures do not normally estimate actual cost changes. This thesis develops an alternative, theoretically defendable method based on linear programming, and tests it on a pine sawmill. Computer simulations are compared with traditional methods, and to analyze the effects of managing the salesmen by the product costs of the suggested method. The thesis relies on the joint-cost accounting discourse from the 1980s, which was abandoned before any essential application was found. This application has now been found through changes in the sawmill industry, and the discourse is here revived practically and theoretically. The sawmills are modeled with relative capacity restrictions and with constraints on the flexibility of their timber supply. Sales decisions based on product costs from the suggested method seem to successively improve company profit. To be successful, the product costs have to be recalculated regularly. Analyses indicate that with flexibility in purchasing timber and a low cost difference between buying scarce products and selling surplus products externally, the necessary length of the recalculation period and the usefulness of the suggested method both increase markedly.