Data with border-line statistical significance, copiously generated in genome-wide association studies of coronary artery disease (CAD), could include functionally relevant associations. We propose an integrated genomic and transcriptomic approach for unravelling new potential genetic signatures of atherosclerosis. Fifteen among 91 single nucleotide polymorphisms (SNPs) were first selected for association in a sex- and age-adjusted model by examining 510 patients with CAD and myocardial infarction and 388 subjects with normal coronary arteries (CAD-free) in the replication stages of a genome-wide association study. We investigated the expression of 71 genes proximal to the 15 tag-SNPs by two subsequent steps of microarray-based mRNA profiling, the former in vascular smooth muscle cell populations, isolated from non-atherosclerotic and atherosclerotic human carotid portions, and the latter in whole carotid specimens. BCL3 and PVRL2, contiguously located on chromosome 19, and ABCA1, extensively investigated before, were found to be differentially expressed. BCL3 and PVRL2 SNPs were genotyped within a second population of CAD patients (n=442) and compared with CAD-free subjects (n=393). The carriership of the BCL3 rs2965169 G allele was more represented among CAD patients and remained independently associated with CAD after adjustment for all the traditional cardiovascular risk factors (odds ratio=1.70 with 95% confidence interval 1.07-2.71), while the BCL3 rs8100239 A allele correlated with metabolic abnormalities. The up-regulation of BCL3 mRNA levels in atherosclerotic tissue samples was consistent with BCL3 protein expression, which was detected by immunostaining in the intima-media of atherosclerotic specimens, but not within non-atherosclerotic ones. Our integrated approach suggests a role for BCL3 in cardiovascular diseases.