Background: Analyses of the urinary proteome have been proposed as a novel approach for early assessment of increased risk of renal- or cardiovascular disease. Here we investigate the potentials of various classifiers derived from urinary proteomics for prediction of renal and cardiovascular comorbidities in patients with type 2-diabetes. Methods: The study was a post hoc analysis of the randomized controlled Steno-2 trial comparing intensified multifactorial intervention to conventional treatment of type 2-diabetes and microalbuminuria. 151 diabetic patients with persistent microalbuminuria were included in year 1995 and followed for up to 19 years. For renal outcomes, two classifiers (CKD273 and a novel, GFR-based classifier) and for cardiovascular outcomes, three classifiers (CAD238, ACSP and ACSP75) were applied. Renal endpoints were progression to macroalbuminuria, impaired renal function (GFR < 45 ml/min/1.73 m2) or progression to end stage renal disease (ESRD) or death. Cardiovascular endpoints were coronary artery disease and a composite endpoint of incident death of cardiovascular disease, myocardial infarction or revascularization, stroke, amputation or peripheral revascularization. Results: CKD273 was not consistently associated with renal outcomes. The GFR-based classifier was associated with impaired renal function, but lost significance in extensively adjusted models. Both the ACSP75 and ACSP-scores, but not the CAD238-score were inversely associated (opposing the hypothesis) with cardiovascular endpoints. None of the classifiers improved prediction of any outcome on top of standard risk factors. Conclusions: Risk-scores based upon urinary proteomics did not improve prediction of renal and cardiovascular endpoints on top of standard risk factors such as age and GFR during long-term (19 years) follow up in patients with type 2-diabetes and microalbuminuria.