Serum levels of lipids, such as cholesterol and triglycerides, are heritable risk factors for cardiovascular disease and targets for therapeutic intervention. Because previous genome-wide association studies (GWASs) did not target functional genetic variants, we employed an alternate approach using nonsynonymous single-nucleotide polymorphisms (SNPs) to identify functional genetic variants associated with the regulation of serum lipid levels. We selected 3667 healthy individuals from a rural community-based cohort (CAVAS; Cardio Vascular disease Association Study) of the Korean Genome and Epidemiology Study project. We analyzed demographic and lifestyle information, lipid measurements and genotypes using the Illumina-1M SNP chip. For genotyping, we isolated 11 558 nonsynonymous SNPs and conducted a linear regression analysis with four lipid traits (total, high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterols and triglycerides). Significantly associated SNPs were validated in two independent Korean populations, Korean Association Resource (KARE) (n=4116) and Health Examinee (HEXA) (n=2178). Of the 11 558 SNPs, one SNP (rs3733197) from the CAVAS was significantly associated with serum LDL cholesterols (beta±s.e.=4.67±0.94, P-value=1.0 × 10(-6 and) Bonferroni corrected P-value=0.012). The replication results of HEXA and KARE were beta±s.e.=2.88±1.12, P-value=0.016 and beta±s.e.=1.26±0.97, P-value=0.196, respectively. An overall meta-analysis of the three data sets revealed beta=2.98±0.57, P-value=6.19 × 10(-7). The rs3733197 is located in the coding region of BANK1 (B-cell scaffold protein with ankyrin repeats 1), and the minor allele (A) resulted in the replacement of the Alanine at position 383 with Threonine.