A total of 1400 soil samples from the plow layer (0-20 cm) at an approximate interval of 5 km were collected in the autumn of 2002 over the entire black soil arable crops region to determine the spatial variability of seven variables, such as total organic matter content (OMC), total N, total P, total K, alkali-dissolvable N (AN), available P (AP) and available K (AK), with classical statistics and geostatistical analysis across the entire black soil area in Northeast China. In nonsampled areas ordinary kriging was utilized for interpolation of estimated nutrient determinations. Classical statistics revealed highly significant (P≤0.01) correlations with all seven of the soil properties, except for OMC with AP and total K with AK. In addition, using coefficients of variation, all soil properties, except for total K, were moderately variable. A geostatistical analysis indicated that structural factors, such as parent material, terrain, and water table, were the main causes of the spatial correlations. Strong spatial correlations were noted with OMC, total N, total P, AN, and AP, while they were moderate for total K and AK. The effective spatial autocorrelation of OMC, total N, total P, and AN ranged from 1 037 to 1 353 km, whereas the ranges of total K, AP, and AK were only from 6 to 138 km. The fit of the experimental semi-variograms to the theoretical models indicated that except for AN, kriging could successfully interpolate other six variables. Thus, the geostatistical method used on a large scale could accurately evaluate the spatial variability of most black soil nutrient properties in Northeast China. A total of 1400 soil samples from the plow layer (0-20 cm) at an approximate interval of 5 km were collected in the autumn of 2002 over the entire black soil arable crops region to determine the spatial variability of seven variables, such as as total organic matter content (OMC), total N, total P, total K, alkali-dissolvable N (AN), available P (AP) and available K (AK), with classical statistics and geostatistical analysis across the entire black soil area in Northeast China. In nonsampled areas ordinary kriging was utilized for interpolation of estimated nutrient determinations. Classical statistics revealed highly significant (P ≦ 0.01) correlations with all seven of the soil properties, except for OMC with AP and total K with AK. In addition, using coefficients of variation, all soil properties, except for total K, were moderately variable. A geostatistical analysis indicated that structural factors, such as parent material, terrain, and water table, were the main causes of the spati al spatial correlations were observed with OMC, total N, total P, AN, and AP, while they were moderate for total K and AK. The effective spatial autocorrelation of OMC, total N, total P, and AN ranged from 1 037 to 1 353 km, and the ranges of total K, AP, and AK were only from 6 to 138 km. The fit of the experimental semi-variograms to the theoretical models that except for AN, kriging could successfully interpolate other six variables Thus, the geostatistical method used on a large scale could accurately evaluate the spatial variability of black soil nutrient properties in Northeast China.