Introduction: Most colorectal cancers are believed to originate from adenomatous polyps that acquire distinct mutations and accumulate other molecular alterations that allow them to progress through distinct histopathologic stages before becoming invasive carcinomas. Adenomatous polyp detection and removal is the ultimate goal of colon cancer screening programs. However not all adenomas have the same potential to progress into carcinomas and thus a test that could detect more advanced adenomas with high sensitivity but could also separate advanced and non-advanced adenomas could be useful for screening purposes. The aim of our study was to show that metabolomic technology can be successfully used to distinguish persons with advanced adenomatous polyps from those without adenomas and the use of the technology could be further used for separation of patients with advanced adenomas from patients with non-advanced adenomas with potential to sub-categorization patients according to their pathological severity. Methods: Prospective serum samples were collected from 131 participants undergoing colonoscopy examination in a colon cancer screening program, from April 2016 to December 2016 at the Hospital Clínic de Barcelona. Colonoscopy reference standard identified 35 participants with no colonic polyps, 12 patients with hyperplastic polyps, 45 patients with non-advanced adenomas and 38 patients with advanced colonic adenomatous polyps. Mass-spectrometry based metabolite and lipid profiles were analysed to define a predictive metabolomic profile for advanced colonic adenomas and confirm if advanced adenoma patients could be separated from non-advanced adenoma patients. Serum metabolomic diagnostic test accuracy was defined for advanced colonic adenomatous polyps using Monte-Carlo 50-fold cross-validation with 70% samples used in a training and 30% of the samples in the validation set. Results: Using random forest algorithm on the metabolomic profile we could identify patients with colonic advanced adenomatous polyps from control patients with 87% sensitivity and 73% specificity. Additionally we could achieve identification of advanced adenoma patients from extended control set that included also patients with hyperplastic polyps and non-advanced adenomas with sensitivity of 81% and specificity of 70%. Additionally we could see clear distinction between advanced and non-advanced adenomas with sensitivity of 80% and specificity of 80%. Classification improvements could be observed even further when classifying adenomas into sub-groups according to severity of the pathology. Conclusion: We describe a proof-of-concept for serum-based metabolomic test that identifies patients with advanced adenomatous polyps with good level of sensitivity (87%) and allows distinction between advanced and non-advanced adenomas with sensitivity of 80% and allows further classification of different adenoma sub-groups according to their pathological severity.