Colorectal carcinoma (CRC) is one of the most common cancers in the world, and approximately 35%-55% of patients with CRC will develop hepatic metastases while their disease. Currently, there are no accurate predictive markers for metastatic CRC. Urine, a kind of promising biofluids, is characterized by its ease of collection, richness in metabolites and its ability to reflect imbalances of all biochemical pathways within the body. To identify stable and reproducible biomarkers for CRC metastasis in noninvasively collected urine, a comprehensive urinary metabonomics analysis of CRC was conducted using the liquid chromatography/high-resolution mass spectrometry (LC-HRMS) strategy. In which, a total 205 urine samples from CRC patients and healthy controls including CRC without metastasis (CRC-NM, n=30), CRC with lymph nodes metastasis (CRC-LNM, n=57) and CRC with liver metastasis (CRC-LM, n=58), as well as healthy controls (HC, n=60) were analyzed at the first stage. Compared to the healthy controls, 40, 50 and 68 kinds of differentiated urinary metabolites were identified in CRC-NM, CRC-LNM and CRC-LM groups. We initially established a urinary metabolomic panel of CRC metastasis containing 4 kinds of urinary metabolites which could distinguish CRC without metastasis and healthy controls both in the training set (Area Under the ROC Curve, AUC=0.938) and in the validation set (AUC=0.898), respectively. Simultaneously, some metabolites could also distinguish CRC-LNM or CRC-LM and HC in both training (AUC=0.882 or 0.937) and validation (AUC=0.818 or 0.882) sets, respectively. The major metabolic pathways of CRC-NM, CRC-LNM and CRC-LM were found including linolenic acid, caffeine and tyrosine metabolism. We explored the urine metabolome by different MS-based techniques and the preliminary findings provided potentially diagnostic urinary biomarkers. The data needs to be validated in large samples sets in the future.