Extracellular vesicles (EVs) are promising biomarkers for cancer diagnosis and prognosis due to their ability to carry specific biomolecular cargo, including DNA. However, the clinical utility of DNA methylation-based liquid biopsies using EV-DNA remains underexplored. The low quantity and relatively long length of EV-DNA complicate whole-genome methylation profiling. To address this, we develop Tn5-assisted Enzymatic Methyl-sequencing with Post-conversion Tailing (TEMPT), a bisulfite-free whole-genome profiling method for EV-DNA. TEMPT employs single-adapter Tn5 tagmentation, enzymatic conversion of unmodified cytosines, and post-conversion tailing to generate high-depth whole-genome EV-DNA methylomes. We apply TEMPT to EV-DNA from 58 gastric cancer and polyp samples, generating methylomes from sub-nanogram inputs and identifying differentially methylated regions (DMRs) that distinguish cancer from controls. We identify potential cancer biomarkers through DMR-associated genes, highlighting the roles of EVs in cellular communication. Our findings suggest that immune cells may serve as an alternative source of EV-DNA. This approach holds significant promise for advancing EV-DNA research and its applications in early disease diagnosis.