Inhibition of glycation by use of anti-glycation agents like aminoguanidine is now regarded as a promising new way for the prevention and treatment of vascular and neuropathic complications in diabetes [1,2]. Several laboratories have reported beneficial effects of aminoguanidine on peripheral nerve function in experimental models of diabetic neuropathy [3±5]. In these studies, the aminoguanidine effects were accompanied by an improvement in nerve blood flow [3,4] and reduction of advanced glycation end product (AGE) formation in nerve tissues [5]. It has also been shown that aminoguanidine treatment can prevent the structural changes of endoneurial microvessels in long-term diabetic rats [6]. The mechanisms for these effects of aminoguanidine and the relation between endothelial cell injury and peripheral nerve dysfunction, however, are not clearly understood.Thrombomodulin, a high-affinity receptor for thrombin, is a membrane glycoprotein found on the surface of endothelial cells [7,8]. Increased plasma Diabetologia (1999) Abstract Aims/hypothesis. Aminoguanidine, a potent anti-glycation reagent, is known to be beneficial in experimental diabetic neuropathy. In this study, we explored the mechanisms of how aminoguanidine inhibits neuropathic changes in diabetes and compared its effects with those of insulin treatment. Methods. Wistar rats, aged 8 weeks, were made diabetic by streptozotocin and given aminoguanidine dissolved in drinking water (1 g/l) for 8 weeks. Effects of daily insulin (protamine-zinc) treatment were also examined for comparison. At the end of the 8 weeks, we examined the peripheral nerve function and (Na + ,K + )-ATPase activity and their relation to serum thrombomodulin concentrations that are considered as a marker of endothelial injury. Results. Aminoguanidine treatment reduced the diabetes-induced decrease in tibial nerve conduction velocity by 47 % (p < 0.05 vs untreated diabetic rats) and inhibited the loss of sciatic nerve (Na + ,K + )-ATPase activity by 54 % (p < 0.05 vs untreated diabetic rats). Insulin-treatment of diabetic rats restored these variables by 83 % and 75 %, respectively (both, p < 0.01 vs untreated diabetic rats). Thrombomodulin concentrations were increased (p < 0.01) in diabetic rats compared with those in non-diabetic controls and unaffected by aminoguanidine treatment. In contrast, the concentrations remained within the normal range in the insulin-treated group. Conclusion/interpretation. Although aminoguanidine treatment improved nerve conduction velocity and (Na + ,K + )-ATPase activity, its effects were considerably less than those of insulin and were not apparent in some measures of endothelial cell injury. [Diabetologia (1999) 42: 743±747]