An integrated approach involving landuse patterns obtained from landuse data and heavy metal contents of the top 10 cm surface soil layer samplings was proposed to detect the characteristic spatial scale of non-point source human disturbances on the Yellow Sea coast in China. Circular plots, with radii of 200. 500, 1000, 1500, and 2000m were set up to represent five spatial scales. We proposed a human impact index (HII) using the landuse data to model the human disturbances. Multivariate statistics of the 10 heavy metals, Cr, Co, Fe, Mn, Ni, V, Zn, Cu. Ti, and Sr, were done. Finally curve estimation between HII and heavy metals was also done. The results showed that: (1) multivariate statistics, including principal component analysis, cluster analysis and the 1-tailed Pearson correlation analysis showing that elements Cr, Cu, Fe, Mn, Ni. V. Zn, and Co could be interpreted as anthropogenic elements and (2) of all the heavy metals showing statistical significance from the curve estimation, in general, the 1000 m scale HII had the best modeling result. We concluded that the characteristic spatial scale of human disturbances on Yellow Sea coast might be 1000 m. Crown Copyright (C) 2010 Published by Elsevier B.V. All rights reserved.