Based on the distribution characteristics of Sb content in each layer of soil, the prediction accuracy and remediation of Sb contamination in each soil layer were analyzed and compared by Kriging interpolation model (OK, UK), inverse distance interpolation (IDW), Voronoi diagram (TIN) and zone prediction model (OK + TIN). The results showed that the order of remediation area determined by each method in 0-1 m soil layer was IDW>OK =UK>TIN>OK+TIN and the remediation area was 74 154, 57 427, 37 338 and 32 707 m² respectively; the order of remediation area determined by each method in 1-3.8 m soil layer was IDW>UK>TIN>OK+TIN>OK, the remediation area was 13 089, 11 030, 9 660, 2 183 and 1 231 m² respectively. The smoothing effect brought by Kriging interpolation resulted in the omission of area exceeding 20 mg/kg and inclusion of area ≤20 mg/kg in the prediction of remediation area. Under the constraints of cost and spatial distribution of pollutants, TIN method could accurately reflect the remediation range of soil Sb within different depths from the perspective of geometric principles; Kriging interpolation including OK+TIN method and IDW interpolation were constrained by the data distribution assumption, with obvious smoothing effect and limited estimation accuracy. The above results have certain guiding significance for the determination of remediation scope of contaminated sites.