Sjögren's Syndrome (SjS) is a chronic systemic immune-mediated inflammatory disease characterized by lymphocytic infiltration and consequent lesion of exocrine glands. SjS diagnosis and classification remains a challenge, especially at SjS onset, when patients may have milder phenotypes of the disease or uncommon presentations. New biomarkers are needed for the classification of SjS, thus, we aimed to evaluate the added-value of lymphocyte subpopulations in discriminating SjS and non-Sjögren Sicca patients. Lymphocyte subsets from 62 SjS and 63 Sicca patients were characterized by flow cytometry. The 2002 AECG and the 2016 ACR/EULAR SjS classification criteria were compared with clinical diagnosis. The added discriminative ability of joining lymphocytic populations to classification criteria was assessed by the area under the Receiver-Operating-Characteristic Curve (AUC). Considering clinical diagnosis as the gold-standard, we obtained an AUC = 0.952 (95% CI: 0.916–0.989) for AECG and an AUC = 0.921 (95% CI: 0.875–0.966) for ACR/EULAR criteria. Adding Tfh and Bm1 subsets to AECG criteria, performance increased, attaining an AUC = 0.985 (95% CI: 0.968–1.000) (p = 0.021). Th1/Breg-like CD24hiCD27+ and switched-memory B-cells maximized the AUC of ACR/EULAR criteria to 0.953 (95% CI: 0.916–0.990) (p = 0.043). Our exploratory study supports the potential use of lymphocyte subpopulations, such as unswitched memory B cells, to improve the performance of classification criteria, since their discriminative ability increases when specific subsets are added to the criteria.