Increased oxygen free radical activity, coupled with reduced protection against oxidative stress, could play a role in the aetiology of neurovascular abnormalities in experimental diabetes mellitus. To test this hypothesis, non-diabetic and streptozotocin-diabetic rats were treated with the anti-oxidant probucol or the pro-oxidant primaquine. One-month diabetes caused 21.4% and 13.6% reduction in sciatic motor and saphenous sensory conduction velocity (p < 0.001). These deficits were prevented by probucol treatment (p < 0.001). After 1-month untreated diabetes, conduction velocity deficits were reversed by a further month of probucol treatment (p < 0.001). For non-diabetic rats, primaquine treatment caused a 12.9% reduction in motor conduction velocity (p < 0.001), which was prevented by probucol treatment (p < 0.001). Primaquine treatment did not affect diabetic rats. Sciatic nerve nutritive endoneurial blood flow, measured using microelectrode polarography and hydrogen clearance, was 48.0% reduced by 2-month diabetes (p < 0.001). This was completely prevented by probucol treatment (p < 0.001). Primaquine treatment did not affect blood flow in diabetic rats. However, in non-diabetic rats it caused a 30.0% reduction (p < 0.01) which was prevented by probucol treatment (p < 0.05). Sciatic endoneurial oxygen tensions were also measured by microelectrode polarography. Mean tension was 38.8% reduced by diabetes (p < 0.001). This was prevented by probucol treatment. Non-diabetic rats given primaquine treatment showed a 21.7% reduction in endoneurial oxygen tension (p < 0.01). The data suggest that vascular-mediated nerve dysfunction in diabetes depends on oxidative stress, and that similar effects in non-diabetic rats may be produced by pro-oxidant treatment. This provides evidence for the potentially important role of oxygen free radical activity in diabetic neuropathy.