Inflammatory bowel disease (IBD) includes ulcerative colitis (UC) and Crohn's disease (CD). Most patients with IBD require a combination of invasive endoscopy, clinical and histopathological examination to confirm the diagnosis. There is a lack of convenient and non-invasive diagnostic methods for IBD. In response to this current situation, a non-invasive diagnostic method for IBD based on surface-enhanced Raman spectroscopy (SERS) was developed. The urine samples from 210 IBD patients (127 CD, 83 UC) and 48 healthy controls (HC) were measured based on SERS. By analyzing the average SERS spectra of urine samples from IBD, CD and HC, it was found that there were significant differences in the Raman intensities at several main Raman peaks between IBD, CD and HC. To validate these findings and construct effective diagnostic models, principal component analysis (PCA)-support vector machine (SVM) was used to establish two classification models to distinguish IBD/HC, CD/HC. After leave-one-patient-out cross-validation (LOPOCV), the accuracy of IBD/HC and CD/HC classification models can reach 83.33% and 83.43%, respectively. The metabolic changes of IBD patients and HC could be effectively identified by measuring urine with SERS, and it was expected to develop a non-invasive, rapid and accurate pre-screening method for clinical IBD patients.