Metabolomics has been applied to explore altered metabolite profiles in disease and identify unique metabolic signatures in recent years. We aim to characterize the metabolic profile of rheumatoid arthritis patients and explore its underlying pathological processes using metabolomics approach. Serum samples from 30 rheumatoid arthritis (RA) patients, 30 primary Sjogren’s syndrome (pSS) patients, and 32 healthy controls (HC) were collected. The sample was analyzed by ultra-high-performance liquid chromatography coupled with high-resolution mass spectrometry (UPLC-HRMS). Potential biomarkers were screened from orthogonal projection to latent structure discriminate analysis (OPLS-DA) and further evaluated by receiver operating characteristic analysis (ROC). Compared with HC and pSS patients, the RA patients had increased serum levels of 4-methoxyphenylacetic acid, glutamic acid, L-leucine, L-phenylalanine, L-tryptophan, L-proline, glyceraldehyde, fumaric acid, and cholesterol as well as decreased capric acid, argininosuccinic acid, and billirubin. A total of eight potential biomarkers were screened and tentatively identified for RA. A panel of three metabolites (4-methoxyphenylacetic acid, L-phenylalanine, and L-leucine) was identified as specific biomarkers of RA. ROC analysis showed that the panel had a sensitivity of 93.30% with a specificity of 95.20% in discrimination RA from other groups. UPLC-HRMS-based quantification of circulating metabolites was a useful tool for identifying RA patients from pSS patients and healthy controls. The potential biomarkers indicated that the RA metabolic disturbance might be associated with inflammation injury, amino acid metabolism, oxidative stress, and phospholipid metabolism.