Conventional soil mapping is generally based on the conceptual models which reflect the relationship between soil genetic factors and its spatial variability,and usually appears to be qualitative and subjective as it is difficult to gather suffi cient data to examine these models over large areas.However,with the development of geostatistics and Geographic Information Systems(GIS),methods of mapping soil properties have also been improved.In this paper,a geostatistic method combined with GIS was applied to characterize and map the spatial variability of total nitrogen(N)in topsoil in the typical area of Taihu Lake watershed.The results showed that the spatial variability of total N were apparent in isotropy within the maximum lag extent of 40 km and its anisotropy had little influence on kriging.The isotropic semivariogram through lags of 40 km indicated that the total N in the studied area had a relatively good autocorrelation,and the range of autocorrelation extended to about 16 km.The maps of the total N made by block kriging based on the fitted exponential model depicted that the soils in Jiangyin and Yixing county gen erally had more N than that in the other two counties and the hot spots of high N were apparent in low-lying polderland.High estimate standard deviations of the kriged estimates occurred where data were sparse and additional samples from the sparse places would increase the accuracy of the total N map which would be useful to conduct optimal fertilization and develop management practices to control the non point sources of N pollution.