Background: We conducted a cohort study to determine if proteinuria predicts cancer-related mortality in type 2 diabetic subjects.Methods: Between July 1996 and June 2003, we enrolled 646 type 2 diabetic subjects. Participants were followed-up until December 31, 2008. The vital status was ascertained by linking records with computerized death certificates in Taiwan.Results: During a median follow-up of 10.4 years, 158 subjects had died, including 59 from cancers. Subjects with proteinuria had a hazard ratio (HR) of 2.77 (95% CI 1.82-4.21) for all-cause mortality and 1.99 (95% CI 1.00-3.94) for cancer-related mortality after adjustment for demographic factors and medical conditions. Specifically, proteinuria showed a trend of increased colon cancer death. The presence of proteinuria significantly improved the predictive ability of cancer-related mortality (increase in concordance statistics or area under the ROC curve = 0.03). Patients with both proteinuria and estimated glomerular filtration rate <60 ml/min per 1.73 m(2) showed higher HR for all-cause mortality than patients with proteinuria only (adjusted HRs (95% CI), 4.01 (2.42-6.67) vs. 2.69 (1.51-4.79), both p<0.01).Conclusions: Proteinuria can predict 10-year all-cause and cancer-related mortalities independently in type 2 diabetic subjects, over and above the established risk factors associated with type 2 diabetes. (C) 2013 Elsevier Inc. All rights reserved.