A 2-year randomized placebo-controlled double-blind trial of 250 mg day-1 of the aldose reductase inhibitor Sorbinil in severe symptomatic chronic peripheral neuropathy was performed in 21 diabetic patients. Due to adverse reactions 8 patients completed the trial on Sorbinil and 6 patients on placebo. Despite differences in baseline neurophysiological parameters, duration of diabetes, and presence of retinopathy between the placebo and treatment groups, there were no initial differences in in vitro platelet aggregation, albumin excretion rate (AER) or muscle capillary basement membrane thickness. After 2 years the median deterioration in AER in the placebo group was 18.4 micrograms min-1 (range 2.6-64.8 micrograms min-1). This deterioration was significant (p less than 0.03). The change in AER in the Sorbinil group was +1.25 (-10.7 to +20.4) micrograms min-1, an insignificant change. In vitro platelet responsiveness to collagen and adenosine diphosphate (ADP) increased in the placebo group (median change max% collagen + 8 (+2 to +30)%; ADP +4.5 (0 to +20)% compared with a fall in the Sorbinil treated patients (median change; collagen -17.5 (-2 to -40)%, p less than 0.05; ADP, -4 (0 to -25)%, p less than 0.05). Muscle capillary basement membrane thickness was measured in only 3 patients in each group and did not alter significantly during the trial. All 12 neurophysiological measurements showed no significant changes between the treatment and placebo groups. The data suggest that aldose reductase inhibition has effects on platelet reactivity and microalbuminuria.