Proteinuria is considered as a predictor for cardiovascular complications in diabetes mellitus (DM). However, no study has examined the association between changes in proteinuria and the risk of diabetic microvascular complications. Study participants were 71,825 DM patients who received urine dipstick test for proteinuria both in 2003-2004 and 2006-2007. They were categorized into four groups according to changes in proteinuria over 3 years (negative: negative → negative, resolved: proteinuria ≥ 1+ → negative, incident: negative → proteinuria ≥ 1+, persistent: proteinuria ≥ 1+ → proteinuria ≥ 1+). Cox-proportional hazard model was used in assessing the adjusted hazard ratios (HR) and 95% confidence interval (CI) for incidence of retinopathy, and neuropathy (adjusted HR [95% CI]). In all of DM patients, risk for comprehensive incidence of retinopathy and neuropathy increased in all types of proteinuria changes. In type 1 DM, HR for retinopathy and neuropathy generally increased in order of negative (reference), resolved (2.175 [1.150-4.114] and 1.335 [0.909-1.961]), incident (2.088 [1.185-3.680] and 1.753 [1.275-2.409]), and persistent proteinuria (1.314 [0.418-4.134] and 2.098 [1.274-3.455]). This pattern of relationship was similarly observed in type 2 DM for retinopathy and neuropathy: negative (reference), resolved (1.490 [1.082-2.051] and 1.164 [0.988-1.371]), incident (1.570 [1.161-2.123] and 1.291 [1.112-1.500]), and persistent proteinuria (2.309 [1.407-3.788] and 1.272 [0.945-1.712]). Risk for diabetic retinopathy and neuropathy generally increased in order of negative, resolved, incident, and persistent proteinuria. Once manifested proteinuria was associated with the increased risk of diabetic retinopathy and neuropathy even after remission of proteinuria.