Phosphorous (P) is an essential nutrient in soil which can lead to eutrophication when eroding into rivers or lakes. As a new method to represent soil properties, fractal theory was applied to study soil characteristics and related process. To understand the correlation of total soil P (TSP) and fractal dimension (D) with the effect of soil depth and slope gradient, soil grid samples were collected at 0-10, 10-20 and 20-40 cm depths in Yingwugou watershed, Yangtze River, China. The physico-chemical properties of the samples were analyzed to determine the TSP content and D. Fractal scaling theory was then applied to establish the relationship between D and TSP. With silt as the dominant particle size, the D values of the study area were in the range of 2.42-2.55. On the flat ground, more fine particles (clay and silt) were reserved and on the contrary relatively high content of coarse sand was retained on steep slope. D values decreased with slope and soil depth increased. The average TSP contents of the three layers were 0.65, 0.63 and 0.57 g/kg for 0-10, 10-20 and 20-40 cm, respectively, showing a decreasing tendency with increased soil depth. With increasing slope gradient, the TSP contents decreased first and then increased slowly. The TSP contents were positively related to silt content (P \ 0.05) and clay content (P \ 0.05) and negatively associated with sand contents (P \ 0.01). There was an one-dimensional linear relationship (n = 530, R 2 = 0.16, P \ 0.01) between TSP content and D values. The correlation would improve when concrete slope was taken into account. To better understand and control soil erosion, the relationship between D value and soil nutrients deserves more attention.