Aldose reductase inhibitors (ARIs) attenuate diabetic complications in several tissues, including lens, retina, kidney, blood vessels, striated muscle and peripheral nerve. However, it is unclear whether their action in diabetes mellitus depends directly on inhibiting the conversion of glucose to sorbitol by aldose reductase or indirectly by reducing the sorbitol available for subsequent metabolism to fructose by sorbitol dehydrogenase. To identify the polyol pathway step most relevant to complications, particularly neuropathy, we compared the biochemical effects of a sorbitol dehydrogenase inhibitor, WAY-135706, (250 mg.kg-1.day-1) and an ARI, WAY-121509, (10 mg.kg-1.day-1) on a variety of tissues, and their effects on nerve perfusion and conduction velocity. After 6 weeks of untreated streptozotocin diabetes, rats were treated for 2 weeks. Sorbitol was elevated 2.1-32.6-fold by diabetes in lens, retina, kidney, aorta, diaphragm, erythrocytes and sciatic nerve; this was further increased (1.6-8.2-fold) by WAY-135706 whereas WAY-121509 caused a marked reduction. Fructose 1.6-8.0-fold elevated by diabetes in tissues other than diaphragm, was reduced by WAY-135706 and WAY-121509, except in the kidney. Motor and sensory nerve conduction velocities were decreased by 20.2 and 13.9%, respectively with diabetes. These deficits were corrected by WAY-121509, but WAY-135706 was completely ineffective. A 48.6% diabetes-induced deficit in sciatic nutritive endoneurial blood flow was corrected by WAY-121509, but was unaltered by WAY-135706. Thus, despite profound sorbitol dehydrogenase inhibition, WAY-135706 had no beneficial effect on nerve function. The data demonstrate that aldose reductase activity, the first step in the polyol pathway, makes a markedly greater contribution to the aetiology of diabetic neurovascular and neurological dysfunction than does the second step involving sorbitol dehydrogenase.