The onset of Sjögren's syndrome (SS) is hidden, early diagnosis is difficult, and the disorder seriously endangers the physical and mental health of affected people. This study aims to identify potential biomarkers of SS and to investigate the characteristics of immune cell infiltration. We used four SS gene expression profile data series from the Gene Expression Omnibus database, and applied bioinformatics analysis and machine learning algorithms to screen two biomarkers, SELL (L-selectin) and IFI44 (interferon-induced protein 44), from 101 differentially expressed genes. The two-gene model comprising SELL and IFI44 showed good diagnostic ability for SS in the training set (AUC = 0.992) and verification set (AUC = 0.917). Analysis of infiltrating immune cells in SS identified naive B cells, resting CD4 memory T cells, activated CD4 memory T cells, gamma delta T cells, M0 macrophages, M1 macrophages, plasma cells, CD8 T cells, activated NK cells and monocytes as candidate participants in the SS process. Furthermore, SELL was associated with M2 macrophages, activated CD4 memory T cells, gamma delta T cells, resting NK cells and plasma cells, while IFI44 was associated with activated mast cells, resting NK cells, resting mast cells and CD8 T cells. This study demonstrates that SELL and IFI44 can serve as good diagnostic markers for SS and may also be new diagnostic and therapeutic targets for SS.