Objective: To systematically determine the serum protein and metabolite expression characteristics, to identify novel biomarkers for disease activity monitoring and clinical diagnoses in patients with systemic lupus erythematosus (SLE). Methods: Serum samples from 121 SLE patients and 106 healthy controls were conducted to proteomics and metabolomics analyses. Disease activity score (SLEDAI) was compared with protein and metabolite expression and clinical data. Random forest machine learning model was performed to identify biomarkers for SLE classification. The clinical utility of the biomarkers was further validated in an independent patient cohort. Results: Screening of the serum proteome and metabolome identified 90 proteins and 76 metabolites significantly changed in SLE patients. Pathway analyses of these molecules revealed SLE related alterations, including immune response, endocytosis and lipid metabolism. Several apolipoproteins and the metabolite arachidonic acid were significantly associated with disease activity. Besides, except some well-known biomarkers, novel molecules such as the protein Apolipoprotein A-IV (APOA4) and the metabolites LysoPC(16:0), punicic acid and stearidonic acid were correlated with renal function in SLE condition. Random forest model by using the significantly changed proteins and metabolites identified 11 proteins and 5 metabolites as potential biomarkers. Among them, 9 proteins (AUC=0.895) and 5 metabolites (AUC=0.902) were validated in an independent patient cohort, which showed good performance for SLE classification.