在引用前人有关土壤侵蚀及其影响因子关系的工作基础上，探讨了神经网络理论在土壤侵蚀估算预报方面的应用。所引用的资料的估算结果表明，用神经网络理论建立的模型估算土壤侵蚀具有很高的精度。同时，文中还简单介绍了神经网络理论及其发展历史。指出了用神经网络理论建立的模型估算土壤侵蚀时应注意的方面。 Based on the previous work on the relationship between soil erosion and its influencing factors, the application of neural network theory in the estimation and prediction of soil erosion is discussed. The estimates of the cited data show that soil erosion is estimated with models built using neural network theory with high accuracy. At the same time, the article also briefly introduces the neural network theory and its development history. The aspects that should be paid attention to when estimating soil erosion with the model established by neural network theory are pointed out.