Background: Human body responses to nutrition and chronic metabolic disorders represent complex features that require integrated systems biology approach to better understand their relationship. Metabolomics, as a powerful phenotyping tool offers now the possibility of characterizing global alterations associated with nutritional exposure and/or disease conditions, and of identifying early and/or predictive biomarkers of disease development. Objective: The objective of this work was to perform a proof of concept study of the use of untargeted metabolomics as part of an integrated strategy for identifying predictive biomarkers of the evolution toward Metabolic Syndrome (MetS), and bring new knowledge about this pathological state. Design: A nested case-control approach was used within the French GAZEL cohort (n~20,000): at risk male subjects (n = 111, 54–64 years old) with high body mass index (BMI, 25 ≤ BMI < 30), free of MetS at baseline, were selected. Cases who developed MetS at the follow-up (5 years later) were compared with Controls (matched for BMI and age) for several parameters (clinical, biochemical parameters, and food habits). Baseline serum samples were analyzed using mass spectrometry-based untargeted metabolomics. A full bioinformatics workflow was optimized to process the data. In particular, data mining methods were used to select the best candidate for prediction. Metabolomic data were then integrated with the different parameters from the database to determine whether multidimensional models improve prediction and impact subject stratification. Results: Metabolomic data allowed improving prediction capacity compared to the use of clinical data only. The multidimensional model integrating metabolomics with anthropometric, biochemical and nutritional data showed the best prediction performances. The analysis of the misclassified subjects revealed sub-phenotypes and generated hypotheses about the disease associated feature space, that need further characterization in order to propose adequate patient sub-stratification and prevention. Conclusion: Metabolomics has the potential to become a key player in systems diagnosis strategies. Integration of metabolic profiles into multidimensional models can be an essential tool for a patient- centered personalized prevention