With the clear physical meaning of parameter structure and the full reflection of spatial heterogeneity,a distributed hydrological model can describe and simulate the real process of runoff induced by rainfall in a basin,accurately and sufficiently.The model is widely needed and concerned recently.Non-point source pollution,land surface process,and climate change impact assessment have a wide range of applications in the simulation of land use,land cover,soil erosion,and other changes in hydrological responses.Accuracy and error are critical to model forecast,which determine the application and extension of the model.On the basis of the analysis of model construction and verification process,the four kinds of model error sources are identified as follows: error caused by the factors being excluded,random or systematic error in a measured historical record,parametric error,and error caused by model structure.The analysis of each error and its calculation method are discussed and a basis for the model development and growth is provided.