Genome-wide association studies with an Illumina Bovine50K chip have detected 105 SNPs associated with one or multiple milk production traits in the Chinese Holstein population. Of these, 38 significant SNPs detected with high confidence by both L1-TDT and MMRA methods were selected to further mine potential key genes affecting milk yield and milk composition. By blasting the flanking sequences of these 38 SNPs with the bovine genome sequence combined with comparative genomics analysis, 26 genes were found to contain or be near to such SNPs. Among them, the gene is merely 87 bp away from the significant SNP, Hapmap30383-BTC-005848. Hence, we report herein genotype-phenotype associations to further validate the genetic effects of the gene. By pooled DNA sequencing of 14 unrelated Holstein sires, a total of 18 with seven novel SNPs were identified. Among them, nine SNPs were in the 5′ regulatory region, one in exon 6 and the other in the 3′ UTR and 3′ regulatory region. A total of nine of these identified SNPs were successfully genotyped and analyzed by mass spectrometry for association with five milk production traits in an independent resource population. The results showed that these SNPs were statistically significant for more than two traits [ <(0.0001-0.0267)]. In addition, mRNA expression analyses revealed that was ubiquitous in eight different tissues, with a relatively higher expression level in the mammary gland than in other tissues. These findings, therefore, provide strong evidence for association of variants with milk yield and milk composition traits and may be applied in Chinese Holstein breeding programs.