Abstract: Spatial variability of soil particle composition was measured by use of geostatistics and GIS techniques for an 87 ha tobacco field with 100 m sampling interval in Chahe, Jia County, Henan Province. The spatial distribution of soil particle and its relationship with soil nutrients were analyzed. The results showed that the semivariogram for soil silt (0.02-0.002 mm) was well fitted by exponential model with medium spatial correlation, while soil clay (<0.002 mm) and sand (0.02-2 mm) by spherical models with medium spatial correlation. The spatial correlation distance for soil silt, clay and sand were 435, 609 and 657 m, respectively. The spatial distribution pattern of soil particle composition was intuitively characterized by kriging interpolation. Soil organic matter, total nitrogen, available phosphorous and available potassium were significantly positively correlated with soil clay content (P<0.01), but negatively correlated with soil sand content (P<0.01). Soil particle composition could be considered as an important index for regionalizing soil management zones