Quantitative proteomics analysis of exosomes isolated from cerebrospinal fluid of neuromyelitis optica patients detected potential distinctive disease markers capable of differentiating from multiple sclerosis and idiopathic longitudinally extensive transverse myelitis. Exosomes were enriched with high purity using differential centrifugation from pooled cerebrospinal fluids. The sample pooling strategy included pre-classification by cytokine level, which improved target molecule identification by providing enhanced fold change ratios and lower p-values when the subsets with high-cytokines were compared to each other. In addition to the spectral counts and label-free quantification, orthogonal quantitative validation was performed using tandem mass spectrometry analysis via UPLC and Orbitrap or Q-Exactive Hybrid Quadrupole-Orbitrap instrumentation. The proteomic datasets with 442 significant exosomal proteins detected putative disease specific biomarkers, glial fibrillary acidic protein and fibronectin for neuromyelitis optica and multiple sclerosis, respectively. Additional investigation of expressions of glial fibrillary acidic protein and fibronectin revealed their presence in intact exosomes as detected by flow cytometry. ELISA measurement of glial fibrillary acidic protein within individual samples demonstrated diagnostic applicability by appropriately clustering disease groups. Expression levels of target molecules from label-free quantification were in good agreement with those from Western blotting. Finally, MetaCore pathway analysis of identified proteins supported the involvement of these proteins in disease progression via neurological pathway involvement. This comprehensive study suggests that the exosomal proteomics approach of cerebrospinal fluid can be applied to the identification and characterization of inflammatory disorders in central nervous system.