Abstract: Statistics and geostatistics methods were adopted to study spatial variations of soil nutrients including total nitrogen (N), available N, available phosphorus (P), available potassium (K), organic matter and pH in top soils from nine counties/cities in main production areas of Jilin province. The results show that the variation coefficients of the soil nutrients are from 10 % to 100 %, which indicate that the variations are at their medium levels. The spatial structures (nugget/sill) of available N, available P available K and organic matter are from 25% and 75%, which indicate moderate spatial self-correlations, while pH and total N are strong spatial self-correlations. According to the decisive coefficients and the least residual sums of squares( RSS) for the model selections, all nutrients are fitted to exponential model well. The GIS-based nutrient spatial variability figures can reflect spatial distributions of the soil nutrients and provide theoretical bases for soil nutrient management and decision-making in the study area.