Aims. – MicroRNAs (miRNAs) from extracellular vesicles (EVs) have been proposed as promising biomarkers for a number of diseases. In this study, their potential as urine-based biomarkers of diabetic nephropathy (DN) was assessed. Methods. – MiRNAs were profiled in urinary EVs from 160 fasting subjects with normal glucose tolerance (NGT) and in T2DM patients with either microalbumininuria (MIC) or macroalbuminuria (MAC). Results. – A total of 73 miRNAs detected in urinary EVs (NGT) were predicted to target important functions for kidney homoeostasis, thereby validating their use as indicators of kidney dysfunction. Indeed, a urinary EV miRNA signature was found to comprise increased levels of let-7i-3p, miR-24-3p and miR-27b-3p, and decreased levels of miR-15b-5p, to identify patients with MIC. ROC curve analysis confirmed this ability to identify MIC in normo-albuminuria T2DM (T2DM-NA) patients and to differentiate between MAC and T2DM patients. These miRNAs were also predicted to target protein networks involved in the Wnt/b-catenin signalling cascade, activin receptor signalling and cell differentiation/proliferation, and correlated with eGRF, HbA1c, serum creatinine, urea, albumin and blood pressure. Concentrations of miR-30a-5p were specifically modified in urinary EVs from patients with MAC, but not MIC, suggesting that miR-30a-5p could be related to severe kidney damage. Conclusion. – Urinary EV miRNAs correlate with the degree of MIC. As they are also thought to regulate pathways that are targets of pharmacological agents to prevent DN (reticulum stress, activin receptors), they may also serve as non-invasive ‘liquid biopsies’ to stratify patients at risk of developing MAC and to monitor treatment efficacy.