Colorectal cancer（CRC） is one of the most vital causes of cancer-related death worldwide，while colorectal adenoma（CA） is an important precancerous lesion of CRC. The development of a CA into a CRC is a relatively long and stealthy process．However，there is still a lack of minimally invasive and reliable biomarkers to distinguish CA from CRC．In this paper，an ultrahigh-performance liquid chromatography－tandem high-resolution mass spectrometry（UHPLC－HRMS） combined with multivariate statistical analysis was used to analyze the untargeted metabolomics of serum samples from 64 CA patients and 84 CRC patients．The acquired metabolomics data were analyzed by unsupervised segregation principal component analysis（PCA） to visualize the grouping trends and detect outliers．A supervised orthogonal partial least squares discriminant analysis（OPLS－DA） was subsequently utilized to maximize the discrimination between the groups．The established OPLS－DA model was validated by 200 times of permutation tests to confirm its rationality and reliability in data analysis．Then the serum differential metabolites responsible for the discrimination between the groups were screened based on the criteria of ,P,<, 0.05 and fold change ,>, 1.50 or ,<, 0.67，and further identified using the mzCloud and HMDB databases．The discriminative ability for differential metabolites was verified by receiver operating characteristic（ROC） analysis．Moreover，the pathway and enrichment analyses of differential metabolites using MetaboAnalyst were used to preliminarily explore the metabolic mechanism of CA cancerization．Results showed that there were certain differences in serum metabolic profiles between the two groups，and 66 differential metabolites were screened and identified，including glycerophospholipids，fatty acids，sphingomyelins，steroids，amino acids，nucleosides and cholines．These differential metabolites mainly involved in the biosynthesis of unsaturated fatty acids，purine metabolism and linoleic acid metabolism，suggesting that they may be closely related to the carcinogenesis of CA．11 metabolic biomarkers including adenine，PC18∶0，arachidonic acid，docosahexaenoic acid，PC 36∶3，8，11，14-eicosatrienoic acid，PC 20∶4，SM d36∶2，5，8，11，14，17-eicosapentaenoic acid，sphingosine（d18∶1），and ACar 20∶1 showed good specificity and sensitivity with the area under the ROC curves greater than 0.80，which had the strong discriminant ability and high potential in clinical application to distinguish CA and CRC．Especially，the area under the ROC curve of the marker panel including PC 36∶3，adenine，sphingosine，PC 18∶0，and PC 20∶4 was 0.941，which presented an outstanding discriminant performance on the CA and CRC．The marker panel discovered provides a valuable reference for the early prevention of CRC clinically.