Modification of dietary fat (1) and carbohydrate (2) have been associated with lower metabolic syndrome risks. This study aimed to examine the changes in metabolic profile of people with higher risk of metabolic syndrome after a dietary intervention. Only samples collected at Imperial College London, one site of the multi-centred RISCK trial (3) were analysed. Subjects (n=70) were initially fed a high saturated fat diet (HS) for four weeks, before randomly assigned to one of the experimental diets for 24 weeks: high monounsaturated fat (MUFA)/high glycemic index (GI) (HM/HGI); high MUFA/low GI (HM/LGI) and low fat/high GI (LF/HGI). The urinary metabolome was measured using proton NMR spectroscopy and analysed with multilevel Partial Least Squares Discriminant Analysis (4). The prediction accuracy of the models was assessed by the difference in correct classification rate between the model and a null model (permutation) using an in-house developed MATLAB script. Figure 1 represents one of the specra obtained, i.e. the HM/HGI group. Overall, the results showed that proline betaine (from citrus), tartaric (from grapes) and glycine (protein sources) increased with the LGI treatment while cis-aconitate increased with the HGI treatment. Additionally, urinary ethanol, metabolites indicative of protein consumption (creatine, carnosine, trimethylamine) and proline betaine could be linked to the LF diet while cis-aconitate and acetaminophen glucuronides could be linked to the HM diet. Application of a data treatment methodology based on a paired data structure has improved the data interpretability and enables the investigation of the effect of dietary modification on human metabolism despite the large between subject variations.