Objective To use Wilks' lambda discriminant analysis for screening of the greatest influence factors in tooth extraction correction and to establish a method for validation.Methods A total of 100 orthodontic patients were selected,aged 14 to 40 years,with average age of 23.6 years,29 males and 76 females.The conventional cephalometry analysis was conducted.Statistical analysis was carried out with using SPSS 18.0 software package.Wilks' lambda statistical method was applied to filter the largest effect on tooth extraction treatment project.Results The UIE-AB and NA/PA calculation results showed statistical differences (P＜0.05),and it set up a tooth extraction rectification regression equation for y =0.125 × NA/PA+0.351 × UIE-AB-3.429;a combination of the initial validation accuracy was 82.6％,and cross validation accuracy was 81.03％.Conclusions Incisor to AB line distance and angle of NA/PA are good tooth extraction judgement indexes.