Objective To discover new biomarkers for diffuse large B-cell lymphoma (DLBCL) diagnosis and prognosis using nuclear magnetic resonance (NMR)-based serum metabolomics. Methods High-resolution serum 1H NMR spectra were collected from 7 DLBCL patients and 7 healthy controls. Spectra were processed using stoichiometry pattern-recognition methods [principal component analysis (PCA) and orthogonal partial least squares discrimination analysis (OPLS-DA)]. Results Significant difference (P <0.05) in 9 metabolites was observed between DLBCL and healthy control by OPLS-DA (variable interpretation rate R2Y ＝ 99.0 % and prediction rate Q2＝ 94.5 %). Compared with the healthy control group, higher levels of lactate (r ＝0.60, P<0.01), glycine (r ＝0.84, P<0.001), creatine (r ＝0.63, P<0.01), and choline (r ＝0.69, P< 0.01), lower levels of acetate (r＝ -0.88, P< 0.001), high-density lipoprotein (r＝ -0.77, P< 0.001), citrate (r ＝-0.82, P<0.001), glutamine (r ＝-0.53, P<0.05) and phosphocholine/glycerophosphocholine (r ＝-0.62, P<0.001) were detected in the serum samples of DLBCL. Conclusion The results of this study offer an evidence for significant changes between DLBCL patients and healthy people in serum metabolite profiles utilizing NMR-based serum metabolomics.