Genome-wide association studies have identified over 300 genomic loci associated with coronary artery disease (CAD) risk, but identifying functional variants remains challenging due to linkage disequilibrium. Here we show a comprehensive functional characterization of CAD-associated variants in primary vascular smooth muscle cells (SMCs). We performed lentivirus-based massively parallel reporter assays (lentiMPRAs) on 25,892 CAD-associated variants, testing their allele-specific enhancer activity in quiescent and proliferative SMCs. We identified 122 candidate variants with enhancer activity and allelic imbalance, including 23 variants showing condition-biased and 41 showing sex-biased effects. Integrating lentiMPRA with CUT&RUN epigenome profiling and expression quantitative trait loci data, we prioritized 49 functionally relevant variants. CRISPRi experiments on eight variants confirmed their regulatory effects on nine variant-gene pairs: rs35976034 (MAP1S), rs4888409 (CFDP1), rs73193808 (MAP3K7CL), rs67631072 (INPP5B/FHL3), rs1651285 (SNHG18), rs17293632 (SMAD3), rs2238792 (ARVCF) and rs4627080 (NRIP3). Our results fine-map the causal variants that confer CAD risk through their effects on vascular SMCs.