Taking the topsoils of the tea-planting area in Karst mountainous regions of Guizhou Province as the study object, based on GIS spatial analysis technology, projection pursuit cluster (PPC) model and potential ecological risk index (RI), as two new evaluation methods, were used to comprehensively evaluate the heavy mental pollution in soil. PPC model takes the contents of Hg, As, Cd, Pb and Cr in 236 soil samples as the multi-dimensional projection parameter to seek its optimal projection direction, obtains the projection value of soil heavy mental pollution by the function of projection index to reflect the characteristic of heavy metal content, establishes the poly class model of projection pursuit, knows the actual situation of soil heavy mental pollution, and further to predict the pollution degree of soil heavy metals and its evolution trend in the study area combined with the method of potential ecological risk index. The results showed that: 1) Hg and As had the greatest influences on soil pollution in the projection direction of index, while Cd had minimal impact on soil pollution. 2) according to the criteria of PPC evaluation, soil environmental quality in the study area was in good condition, the areas of no-pollution level accounted for 96.61% of the study area, the numbers of clean level and still clean level accounted for 30.08% and 66.53% of the total samples, respectively. 3) There were low level and moderate level in the potential ecological risk, accounted for 89.41% and 10.59% of the total samples, respectively, their risk degree were low, but trended to worsen into moderate level. 4) On pollution level, the clean level and still clean level distributed widely in the study area, the spatial distribution of light pollution level was not obviously after interpolation. The above results are favorable for the comprehensive evaluation of soil environmental quality and can play a reference role in the management of soil resources and the rational distribution of tea-planting in Karst areas.