In this study, we verified the hypothesis that Raman signature of urinary extracellular vesicles (UEVs) can be used to stratify patients with diabetes at various stages of chronic kidney disease (CKD). Patients with type 2 diabetes diagnosed with different stages of CKD and healthy subjects were enrolled in the study. UEVs were isolated using low-vacuum filtration followed by ultracentrifugation. Correlation analysis, multiple linear regression and principal component analysis were used to find differences between spectral fingerprints of UEVs derived from both groups of patients. Electron microscopy and nanoparticle tracking analysis were applied to characterize the size and morphology of UEVs. We observed significant correlations between selected Raman bands measured for UEVs and clinical parameters. We found significant differences in the area under the specific bands originating mainly from proteins and lipids between the study groups. Based on the tryptophan and amide III bands, we were able to predict the estimated glomerular filtration rate (eGFR). Principal component analysis, partial least squares regression (PLSR) and correlation analysis of the UEV Raman spectra supported the results obtained from the direct analysis of Raman spectra. Our analysis revealed that PLSR and a regression model including tryptophan and amide III bands allows to estimate the value of eGFR.