BACKGROUND: The causes and mechanisms of increased mortality of patients with diabetic nephropathy are unclear, and its natural history is poorly understood. Aim: To evaluate risk factors for mortality in type 2 diabetic patients with nephropathy. DESIGN: Retrospective study of clinical and biochemical parameters in diabetic nephropathic patients and controls sampled from a secondary care register. METHODS: We studied 170 type 2 diabetic patients (from 1987 to 1995) with nephropathy (proteinuria >0.5 g/24 h) and 170 non-nephropathic patients. Follow-up was until death or December 1997. Details of demographics, clinical and treatment history were obtained from medical records. RESULTS: Mean follow-up was 5.3 years. Of the patients with nephropathy at baseline, 63 (37%) died compared with 14 (8%) non-nephropathic patients (chi(2)=53.8, p2 g/24 h, 9.9 (7.3-13.5) per 100 patient-years. A 36% (5-78%) excess risk of mortality was observed for each log unit increase in proteinuria. Multivariate Cox regression analyses confirmed a five-fold excess risk for all-cause and cardiovascular mortality in patients with nephropathy compared with those without. This was independent of other risk factors including baseline age [5% (1-8%)/year], creatinine [2.5 (1.12-5.6)/10 micromol/l] and glycaemic control (HbA(1c)) [15% (1-31%) per 1% rise]. CONCLUSIONS: Proteinuria is a potentially preventable and reversible risk factor associated with high mortality in type 2 diabetic patients. Prevention of the development of overt nephropathy and improvement in diabetes control may reduce mortality in these patients.