Based on 194 soil samples collected from the studied zone in the suburbs of Suzhou,a minimum dataset(MDS) for soil quality assessment was defined and developed with the aid of multivariate statistical techniques,such as multivariate analysis,principal component analysis and correlation analysis.Different from related previous work,the study took into account effects of the pattern and duration of land-use on soil quality as an index of MDS,and worked out integrated load of each candidate soil physico-chemical variable on all principal components with eigenvalue >1 rather than just compare the load of a variable on a principal component and then rule out the others,thus avoiding losing too much soil quality information.Finally,reliability of the method was validated from different angles by applying the method to 54 Huangnitu samples,77 Huangsongtu samples and all the 194 samples.The same MDS produced suggestes that the method is very good in stability and reproducibility.Once the method is improved,it could be extrapolated to other soil types in other areas of China,so as to overcome the shortcoming of uncomparability of the results of soil quality assessment using different methods because the soils vary so much in geographic distribution and natural conditions.