Background Sarcoidosis features non-necrotizing granulomas consisting mainly of activated CD4-lymphocytes. T-cell activation is regulated by immune checkpoint (IC) molecules. The present study aimed to compare IC expression on CD4, CD8 and NK cells from peripheral, alveolar and lung-draining lymph node (LLN) samples of sarcoidosis patients. Methods Flow-cytometry analysis was performed to detect IC molecules and a regression decision tree model was constructed to investigate potential binary classifiers for sarcoidosis diagnosis as well as for the IC distribution. Results Fourteen patients (7 females) were consecutively recruited in the study; all enrolled patients showed hilo-mediastinal lymph node enlargement and lung parenchyma involvement with chest X-rays and high resolution computed tomography. CD4+PD1+ and CD8+PD1+ were higher in bronchoalveolar lavage (BAL) than in LLN (p = 0.0159 and p = 0.0439, respectively). CD4+ T-cell immunoglobulin and ITIM domain (TIGIT)+ were higher in BAL than in peripheral blood mononuclear cells (PBMCs) (p = 0.0239), while CD8+TIGIT+ were higher in PBMC than in BAL (p = 0.0386). CD56+TIGIT+ were higher in LLN than in PBMC (p = 0.0126). The decision-tree model showed the best clustering cells of PBMC, BAL and LLN: CD56, CD4/CD8 and CD4+TIGIT+ cells. Considering patients and controls, the best subset was CD4+CTLA-4+. Conclusion High expression of PD1 and TIGIT on T cells in BAL, as well as CTLA-4 and TIGIT on T cells in LLN, suggest that inhibition of these molecules could be a therapeutic strategy for avoiding the development of chronic inflammation and tissue damage in sarcoidosis patients. Miriana d'Alessandro & Laura Bergantini contributed equally to the study.