There are several notable challenges inherent for fully characterizing the entirety of the human saliva proteome using bottom-up approaches, including polymorphic isoforms, PTMs, unique splice variants, deletions, and truncations. To address these challenges, we have developed a top-down based LC-MS/MS approach, which cataloged 20 major human salivary proteins with a total of 83 proteoforms, containing a broad range of PTMs. Among these proteins, several previously reported disease biomarker proteins were identified at the intact protein level, such as beta-2 microglobulin. In addition, intact glycosylated proteoforms of several saliva proteins were also characterized, including intact N-glycosylated protein prolactin inducible protein and O-glycosylated acidic protein rich protein. These characterized proteoforms constitute an intact saliva proteoform database, which was used for quantitative comparison of intact salivary proteoforms among six healthy individuals. Human parotid and submandibular/sublingual gland secretion samples (2 μg of protein each) from six healthy individuals were compared using RPLC coupled with the 12T FT-ICR mass spectrometer. Significantly different proteoform profiles were resolved with high reproducibility between parotid secretion and submandibular/sublingual glands. The results from this study provide further insight into the potential mechanisms of PTM pathways in oral glandular secretion, expanding our knowledge of this complex yet easily accessible fluid. Intact protein LC-MS approach presented herein can potentially be applied for rapid and accurate identification of biomarkers from only a few microliters of human glandular saliva.