The level of erythrocyte aldose reductase protein (AR-p) was determined in diabetic patients as well as in 76 healthy controls by a two-site enzyme-linked immunosorbent assay. No significant difference in the mean AR-p level was demonstrated between the healthy and diabetic individuals. Based on the results of seven nerve function tests, 95 non-insulin-dependent diabetes mellitus (NIDDM) patients were classified into two groups: Group I, without demonstrable neuropathy (≤1 abnormal test results); Group II, overt neuropathy (≥2 abnormal results). The AR-p level was significantly higher in Group II than that in Group I. Multivariate logistic regression analysis identified two independent risk factors for overt neuropathy: longer duration of diabetes after clinical diagnosis (odds ratio, 1.15 per year; 95% confidence interval, 1.05–1.25) and a higher level of AR-p (odds ratio, 1.92 per 1 ng mgHb⁻¹; 95% confidence interval, 1.39–2.65). On 31 patients the AR-p level was re-assessed after a 12-month follow-up period. Irrespective of improved or stable HbA_1c levels during the follow-up period, no apparent alteration in the level of AR-p was demonstrated. These results suggest that erythrocyte AR-p level may affect the susceptibility or development of diabetic neuropathy in NIDDM patients.