Abstract: To quantitatively describe and analyze the structural composition and spatial configuration of forest landscape is a key approach to effectively realize the link between the forest landscape structure and function. Based on the calculation and analysis on the regional landscape indices with FRAGSTATS, and by using BP artificial neural network as the tool for modeling construction, a response analysis was conducted on the forest landscape pattern to topography, water area, and human activities in Loufan County of Shanxi Province, China. In the study area, the landscape diversity was relatively low, and the bare hill and wasteland for afforestation as well as the shrub land dominated the landscape element composition. The homogeneity index of the landscape in the study area was relatively high, and there was a relative even distribution of the landscape patches among the landscape element types. The BP artificial neural network model could well simulate the response relationships between the landscape indices and their formation factors, with a high fitting precision and small error. When the ecological factors such as residents’ area, elevation, and distances to river and lake water were known, the forest landscape diversity and fractal dimension could be quickly and easily simulated by the BP artificial neural network model. This paper provided a new perspective to study the management, rational use, and conservation of forest resources.