Extracellular vesicles (EVs) derived from pleural effusion (PE) is emerging as disease biomarkers. However, the methods for isolation of EVs from PE (pEVs) were rarely studied. In our study, three methods for isolating pEVs of lung cancer patients were compared, including ultracentrifugation (UC), a combination of UC and size exclusion chromatography (UC-SEC) and a combination of UC and density gradient ultracentrifugation (UC-DGU). The subpopulation of pEVs was identified by nanoparticle tracking analysis (NTA), transmission electron microscopy (TEM), Western blotting (WB) and nanoflow cytometry (nFCM). Additionally, the proteomic landscape of pEVs was analyzed by Label-free proteomics. The results showed that compared with UC and UC-DGU, the UC-SEC method separated pEVs with the highest purity. In the proteomic analysis, on average, 1595 proteins were identified in the pEVs isolated by UC-SEC, much more than pEVs isolated by UC (1222) or UC-DGU (807). Furthermore, approximately 90% of identified proteins in each method were found in the EVs public database ExoCarta. Consistent with this, GO annotation indicated that the core proteins identified in each method were mainly enriched in “extracellular exosome”. Many of the top 100 proteins with high expression in each method were suggested as protein markers to validate the presence of EVs in the MISEV2018 guidelines. In addition, combined with lung tissue-specific proteins and vesicular membrane proteins, we screened out and validated several novel protein markers (CD11C, HLA DPA1 and HLA DRB1), which were enriched in pEVs rather than in plasma EVs. In conclusion, our study shows that the method of UC-SEC could significantly improve the purity of EVs and the performance of mass spectrometry-based proteomic profiling in analyzing pEVs. The exosomal proteins CD11C, HLA DPA1 and HLA DRB1 may act as potential markers of pEVs. The proteomic analysis of pEVs provides important information and new ideas for studying diseases complicated with PE.