The overall objective of the project is to develop accurate and robust markers of the evolution of health status toward metabolic syndrome (MetS), and to determine to what extent nutrition is a major determinant, by using a multidisciplinary approach, putting together sociology, epidemiology, nutrition, statistics, and computer science. The project uses the French population-based cohort GAZEL, an on-going epidemiological study set up in 1989 (around 20,000 volunteers) among employees of the French national Gas and Electricity Company. The study consists in integrating demographic, socioeconomic, clinical, and biological data (from annual questionnaires, including food frequency questionnaires (FFQs)) to analyse food trajectories between 1998 and 2009. A case-control approach is used within a sub-cohort to characterize metabolic signatures and identify early discriminant factors predictive of MetS development. Male subjects between 52 and 64 years old, with high BMI (25≤BMI<30) who developed MetS (NCEP criteria) after the follow-up (‘Case’ group) were selected and compared for several parameters (socio-demographic, collected clinical, biochemical parameters, and food habits) with those who did not but still having some risk factors (‘Control’ group, matched for BMI, age and sex). Metabolomic analyses of serum samples collected between 2000 and 2003 will be performed using a mass spectrometry-based untargeted approach. Analyses of correlations between social characteristics, food habits and metabolic signatures will be done to build predictive models and determine whether integration of multidimensional parameters improves prediction. Among the 9,000 individuals having responded to FFQs in 1998, three different dietary patterns were identified, Western Diet (WD), Healthy, and Traditional French, after Multiple Correspondence Analyses (MCA) based on 22 food items. These patterns were confirmed from FFQs in 2004 and 2009. Food trajectories of individuals were found to be different according to their dietary patterns: strong decrease of ‘WD’ dietary pattern with time, in opposition with an increase in ‘Healthy’ and ‘Traditional’ behaviours. The at-risk sub-cohort appears to be representative of the whole cohort after MCA analyses on the 22 identified food variables, the nature of the dietary patterns, and the trend of changes. However, the evolution of behaviours for the Cases (MetS) and Controls is different in its amplitude: a deeper decrease in ‘WD’ pattern in the Cases in favour of an increase in ‘Traditional’ behaviour. These data will be correlated with metabolic changes identified from metabolomic profiles. This approach should provide new tools to better stratify at-risk populations. Furthermore, these analyses will allow identifying the role of nutrition as determinant and modulator in MetS etiology. Finally, this study should contribute to develop a more personalized nutritional prevention.