Using 54,602 SNPs, a genome-wide association study (GWAS) was performed to identify genomic regions associated with estimated breeding values (EBV) from racing times, as determined by the Korea Racing Association (KRA). The heritability and repeatability were estimated at 31% and 43 % respectively based on the quantitative trait model, suggesting that racing performance is heritable. Following GWAS with SNPs, we identified significantly associated 39 SNPs with EBV for racing time using simple linear regression methods. Candidate SNPs contained 23 genes annotated by ENSEMBL. These findings suggest the presence of genes affecting the racing performance in Thoroughbred racehorses covering entire most chromosomes. From associated genes with racing time we confirmed 5 genes are reported relationship with racing performance in previous research. We suggest that our procedure of association study for complex and repeated trait is applicable and candidate genes associated with racing performances in thoroughbred.