Studies with the aldose reductase inhibitor alrestatin in animal models have suggested that the sorbitol pathway may be of etiologic significance in the pathogenesis of peripheral neuropathy in diabetes. In normal subjects and in highly selected diabetic patients with severe peripheral neuropathy, alrestatin given either intravenously (50 mg/kg body weight) or orally (1 gm q.i.d.) produced no acute toxicity. The serum half-life of alrestatin was approximately 1 hr, and 99% was recovered in the urine within 24 hr. Two diabetic patients receiving alrestatin intravenously reported subjective improvements in clinical symptoms 2 days following the start of infusions. These improvements lasted approximately 3 wk after infusions were discontinued. However, there were no significant objective changes in peripheral nerve condition velocities, or on neurologic examination. In a 30-day oral trial with alrestatin in 4 diabetics, there were no subjective improvements in clinical symptoms nor were there objective improvements on neurologic examination or in peripheral nerve conduction velocities. In this study, peak serum levels of alrestatin were approximately 3 times lower than those obtained on intravenous administration, and it is possible that a high peak serum level is critical to the attainment of adequate tissue drug concentrations. Furthermore, the patients were suffering from severe clinical peripheral neuropathy, which could represent a stage of permanent irreversible nerve damage. Studies with alrestatin in newly diagnosed diabetics with peripheral nerve conduction velocity deficits but without clinical neuropathy might provide a better test of the sorbitol pathway hypothesis.