Background: Previous genome-wide association studies (GWAS) have identified several risk genes for stroke; however, it remains unclear how they confer risk for the disease. We conducted an integrative analysis to identify candidate genes for stroke and stroke subtypes by integrating blood-derived multi-omics data with genetic data. Method: We systematically integrated the latest stroke GWAS database (73,652 patients and 1,234,808 controls) with human plasma proteomes (N=7,213) and performed proteome-wide association studies (PWAS), Mendelian randomization (MR), Bayesian colocalization analysis, and transcriptome-wide association study (TWAS) to prioritize genes that associate the risk of stroke and its subtypes with their expression and protein abundance in plasma. Cell-type specificity and functional enrichment analysis using single-cell RNA sequencing (scRNA-seq) and Gene Ontology (GO) databases were then performed to select target genes. A two-step MR analysis was followed to explore the potential mechanisms. Results: We found that the protein abundance of seven genes ( MMP12, F11, SH3BGRL3, ENGASE, SCARA5, SWAP70, and SPATA20 ) in the plasma was associated with stroke and its subtypes, with six genes ( MMP12, F11, SH3BGRL3, SCARA5, SWAP70, and SPATA20 ) causally related with stroke and its subtypes (P < 0.05/proteins identified for PWAS; P < 0.05/8 for MR; posterior probability of hypothesis 4 ≥ 75 % for Bayesian colocalization). The effect of F11, SH3BGRL, SPATA20, and SWAP70 on each subtype was mediated by Factor XI inhibitors (FXI), atrial fibrillation, T2D, and SBP respectively (p<0.05). We also found that SCARA5 and SWAP70 were related to stroke and ischemic stroke at the transcriptome level. Conclusions: Our present proteomic findings have identified new causal genes in the pathogenesis of stroke, which may offer potential future therapeutic targets for stroke prevention.