Abstract: 【Objective】 Harvest adjacency and green-up constraints have become the most commonly used constraint types for spatial forest harvest scheduling in forestry in developed countries worldwide in the last decade; however, few studies have focused on this issue in our country. Therefore, the concept of this forest-management technique will be discussed thoroughly with an example from Northeast China, which can provide certain insights into the sustainable management of forest ecosystems in our country. 【Method】 Using a simulated annealing algorithm as an optimization technique, three increasingly difficult forest-planning problems were analyzed for the Pangu Forest Farm in the Da Hinggan Ling Prefecture, Northeast China, which were used for analyzing the effects of different spatial constraint types on the results of forest planning. The objective functions for all three planning problems were to maximize the discounted net present value of forest ecosystems for timber production. The first problem, a non-spatial problem, did not include any spatial information. However, the second and third problems, i.e., the unit restriction model(URM)and area restriction model(ARM)problems, were analyzed based on the non-spatial problem. The URM problem strictly prohibited the scheduling of neighboring management units for a final harvest during the same period; however ARM allowed for the scheduling of certain limited neighboring units for final harvest during the same period, provided that the total final-harvest area was less than the user-defined maximum size. Additionally, all three planning problems were subjected to the even-flow harvest constraint, green-up constraint, minimum harvest-age constraint and number of harvest constraints for each unit(or stand). 【Result】 The results showed that the coefficient of variation of the objective function values for each planning problem only ranged from 0.06% to 3.97%, indicating the great stability of the simulated annealing algorithm. The objective function values of the ARM problems increased slightly(0.08%),although not significantly(P=0.35), and those of the URM problems reduced significantly(5.11%, P<0.01),compared with those of the non-spatial problem; however, the temporal and spatial outputs of the forest management treatment became more reasonable. The percentage of all harvest areas across the planning horizon of the optimal forest-management plan for each planning period was relative less, only accounting for approximately 0.44% of the total area in the forest dataset. These results were reasonably and perfectly consistent with the criteria of sustainable forest management. 【Conclusion】 The spatial constraints for forest-management treatments increase the complexity of forest-planning models, and significantly decrease the economic benefits of timber production from forest ecosystems, as well; however, the outputs of forest-management plans might be more suitable for sustainable forest management. Additionally, most heuristic techniques, including simulated annealing algorithms, can be applied to larger and more complex forest-planning problems, as performed in this study.