To investigate the association of polymorphisms of PAX1 gene with congenital scoliosis (CS) in Chinese Han population and the relationship between the PAX1 gene polymorphisms and the clinical phenotypes of CS. Peripheral blood samples were collected from 127 CS patients, 55 male and 72 female, aged (12.9 +/- 4.3) (2 - 23), and 127 sex- and age-matched controls. Based on genotype data from the International HapMap project, the tagging single nucleotide polymorphisms (tSNPs) were selected using Haploview 4.0 software. Hardy-Weinberg equilibrium was analyzed both in the control and case groups. The case group was classified into different clinical phenotypes according to vertebral defect type, location of deformity, extent of developmental disruption, combined rib malformations, and neural canal deformity. Genotyping of all selected SNPs was done by SNPstream technology. The association between phenotypes and SNP was analyzed. Pairwise linkage disequilibrium was calculated in the control population using Haploview 4.0 software. The sites: SNP1 (rs17861031) and SNP2 (rs6047590), of PAX1 gene were genotyped and both polymorphisms were distributed in line with the Hardy-Weinberg equilibrium in these 2 groups. There was no linkage disequilibrium between these 2 SNPs. The genotype frequencies of SNP1AA, SNP1AG, SNP1GG, SNP2AA, SNP2AT, and SNP2TT of the case group were 2%, 26%, 72%, 2%, 19%, and 80% respectively, all not significantly different from those of the control group (2%, 26%, 72%, 2%, 26%, and 82% respectively, all P > 0.05). The allele frequencies of SNP1A, SNP1G, SNP2A, and SNP2T of the case group were 15%, 85%, 11%, and 89% respectively, all not significantly different from those of the control group (15%, 85%, 10%, and 90% respectively, all P > 0.05). No positive sites were found in different clinical phenotypes. The genetic variants of PAX1 gene may not be associated with the susceptibility to CS and different clinical phenotypes of CS in Chinese Han population.