The cost efficiency of any large-scale, extensive forest inventory is decisively influenced by the sampling design. Many different sampling techniques have been suggested, e.g. double sampling for stratification and sampling with partial replacement. This study presents models for optimization which can help in selecting the optimum design from a range of alternatives. The selection of the optimum alternative is usually based on cost efficiency. The aim of optimization is either to minimize costs for a specified precision or to minimize the variance given a fixed cost. For each of these optimization rules models increasing transparency and objectivity in selecting a design have been developed.