A neural network is an information processing technique based on knowledge about such $quot;biological computers$quot; as the human brain. The highly-interconnected network of nonlinear analog neurons rapidly provide a collectively-computed solution to a problem on the basis of input information. The computational effectiveness of neural networks has the potential to sole a wide variety of forestry decision-making problems. This research investigated the computational efficiency and accuracy of the neural network for inferring optimal stand management regimes from initial stand conditions. Using forest stand management case studies, the performance of neural networks were examined in terms of size of training set, structural aspects of the networks and the degree of network learning. For experimental purposes, two different non - linear programming stand optimization models were used to generate training and verification sets for the neural network analysis. The two models were designed to optimize stand management prescriptions fur longleaf pine (Pines palustris) stands and mixed hardwood stands, respectively. Networks were built and trained to recognize optimal solution patterns associated with current stand stocking parameters. A backpropagation algorithm was used as the learning rule of the neural network .