We present a hierarchical approach for large-scale forest planning. It is based on solving an aggregate problem, which is of moderate size, at higher levels of derision. The results obtained lead to land allocations and input-output targets for the different zones of the forest. At a lower level of decision, based on the land allocation and input-output targets, each zone is modeled in a detailed way. These problems are again of moderate size. Consistency between the two levels of derision is preserved. The economic and algorithmic implications of this methodology approach are discussed. A test case is presented to show the feasibility of this approach. For. Sci 37(2):439-460.