Background: A close relationship exists between major depressive disorder (MDD) and diabetes mellitus. The metabolomic difference and similarity between patients with and without diabetes mellitus have not been well studied in the context of MDD. We aimed to examine these differences and common serum metabolomics patterns, pathways and biomarkers that can comprehensively reflect the pathogenetic difference and similarity between these MDD groups. Methods: We performed a metabolomics analysis of serum samples of healthy controls (n = 6), patients with MDD and type 2 diabetes mellitus (n = 13), and patients with MDD without type 2 diabetes mellitus (n = 27). Metabolomics analysis was conducted using capillary electrophoresis Fourier transform mass spectrometry and a candidate compound was assigned to the 496 (290 cation, 206 anion) peaks. Moreover, we evaluated the sensitivity and specificity of the candidate biomarkers for distinguishing between MDD patients with or without type 2 diabetes mellitus. Results: Principal component analysis revealed no clear distinction among the three groups, while naive partial least squares discriminant analysis yielded three relatively good and distinct populations based on the first principal component. Energy conversion by the tricarboxylic acid cycle represented the highest percentage among the top 30 positive factors of the first principal component, and glutamate metabolism and urea cycle represented the highest percentage among the top 30 negative factors of the first principal component. Synthesis and degradation of ketone bodies had high impact in MDD with type 2 diabetes mellitus group and taurine and hypotaurine metabolism had high impact in MDD without type 2 diabetes mellitus group for the pathway. Conclusions: Patterns of serum metabolites may be different among MDD with type 2 diabetes mellitus, MDD without type 2 diabetes mellitus, and healthy controls groups. Specifically, comorbid type 2 diabetes mellitus could affect metabolomics pathway and alter the distribution of serum metabolites in patients with MDD. These findings may shed light on the influence of the type 2 diabetes on the pathophysiology of MDD.