OBJECTIVE To evaluate putative risk factors for the development of incipient diabetic nephropathy (persistent microalbuminuria) and overt diabetic nephropathy (persistent macroalbuminuria) in patients with non-insulin dependent diabetes. METHODS Prospective, observational study of a cohort of white, non-insulin dependent diabetic patients followed for a median period of 5.8 years. METHODS Outpatient clinic in tertiary referral centre. METHODS 191 patients aged under 66 years with non-insulin dependent diabetes and normoalbuminuria (urinary albumin excretion rate < 30 mg/24 h) who attended the clinic during 1987. METHODS Incipient and overt diabetic nephropathy. RESULTS Fifteen patients were lost to follow up. Thirty six of the 176 remaining developed persistent microalbuminuria (30-299 mg/24 h in two out of three consecutive 24 hour urine collections) and five developed persistent macroalbuminuria (> or = mg/24 h in two out of three consecutive collections) during follow up. The five year cumulative incidence of incipient diabetic nephropathy was 23% (95% confidence interval 17% to 30%). Cox's multiple stepwise regression analysis revealed the following risk factors for the development of incipient or overt diabetic nephropathy: increased baseline log urinary albumin excretion rate (relative risk 11.1 (3.4 to 35.9); P < 0.0001); male sex (2.6 (1.2 to 5.4); P < 0.02); presence of retinopathy (2.4 (1.3 to 4.7); P < 0.01); increased serum cholesterol concentration (1.4 (1.1 to 1.7); P < 0.01); haemoglobin A1c concentration (1.2 (1.0 to 1.4); P < 0.05); and age (1.07 (1.02 to 1.12); P < 0.01). Known duration of diabetes, body mass index, arterial blood pressure, serum creatinine concentration, pre-existing coronary heart disease, and history of smoking were not risk factors. CONCLUSIONS Several potentially modifiable risk factors predict the development of incipient and overt diabetic nephropathy in normoalbuminuric patients with non-insulin dependent diabetes.