Because of environmental concerns, modern forest harvesting plans are being modified to address aesthetics and wildlife protection issues. In the literature, the most referenced are adjacency constraints with exclusion or “greenup” years. Also of interest are the concepts of old growth patch size and total old growth area restrictions. Typically, harvest blocks have been defined a priori. Recently, some have proposed a more complex approach in which basic cells are defined, and the decision-making process includes forming harvest blocks from these cells. This has been called ARM (Area Restriction Model). Until now, no successful exact method has been reported for large-scale ARM problems, leaving heuristics as a viable option. In this work we present a multiperiod harvest-scheduling ARM model with these spatial constraints. A tabu search procedure with 2-Opt moves (exchanging at most 2 units) was developed. For small size instances, the procedure found the optimal solution in most cases, and for a real-size problem an 8% optimality gap is provided.