Introduction Stroke is a leading cause of death and disability worldwide. Approximately one quarter of all strokes are secondary to carotid atherosclerosis. There is a clinical need to improve risk stratification of carotid atherosclerosis, to better target surgical or interventional therapy and prevent stroke. This study aimed to determine diagnostic biomarkers of high-risk carotid atherosclerosis, and ensure the validity of such markers in the presence of alternative phenotypes of atherosclerotic disease. Methods 150 patients were recruited according to the following criteria: Group 1: Symptomatic > 50% carotid stenosis Group 2: Non-carotid stroke/TIA Group 3: Asymptomatic >50% carotid stenosis Group 4: Asymptomatic controls with < 50% carotid stenosis Group 5: Abdominal aortic aneurysm Group 6: Intermittent claudication Disease groups were matched for age, gender, cardiovascular risk factors, haematological parameters, renal function and lipid status. Blood and urine was collected from all patients and analysed through global metabolic profiling (1H-NMR Spectroscopy, HILIC-Mass Spectrometry and Lipid Profiling-Mass Spectrometry). Acquired spectra were compared across groups using computational multivariate data analysis to determine markers of high-risk carotid atherosclerosis. Results Statistical models derived from urinary spectra proved stronger than serum datasets, in particular with HILIC-Mass Spectrometry (positive ionisation mode). Application of computational OPLS DA resulted in discrimination of symptomatic carotid atherosclerosis from asymptomatic disease, aneurysmal disease, and intermittent claudication. Differentiating metabolites span a vast array of compounds including lipid derivatives, amino acid derivatives and nucleotide derivatives. Conclusion This is the first study to identify urinary metabolic biomarkers of high-risk carotid atherosclerosis, differentiating symptomatic carotid atherosclerosis from asymptomatic disease, and aneurysmal and peripheral arterial disease. Targeted temporal studies are now required for clinical validation and to determine the variation of acute biomarkers with time.