Analysis of metabolic biomarkers for Type 2 diabetes (T2D) can shed insight into etiological pathways and improve the prediction of disease risk. Here, we aimed to identify serum metabolites as potentially informative biomarkers for T2D. We quantified the production of 123 metabolites from 2,240 subjects in the Korea Association Resource (KARE) cohort. Four metabolites, hexadecanoylcarnitine (C16), glycine, lysophosphatidylcholine acyl C18:2 (lysoPC a C18:2), and phosphatidylcholine acyl-alkyl C36:0 (PC ae C36:0), were significantly altered in T2D compared to non-T2D subjects, with P value < 4.07E-04 (after Bonferroni’s correction for multiple testing, α = 0.05). Among them, C16, glycine, and lysoPC a C18:2 were independently replicated in the population-based Cooperative Health Research in the Region of Augsburg (KORA) cohort. In addition, by combining a genome wide association study (GWAS) with metabolomics, alterations of these metabolites were associated with 10 genetic loci including six that were previously implicated in T2D or obesity. These results may help in developing novel strategies for preventing T2D in the Korean population.