Previous studies indicate that experimental diabetic autonomic neuropathy can be largely prevented by initiating therapy at the onset of diabetes. More clinically relevant, however, is the ability of therapy to reverse established neuropathy produced by longstanding diabetes. We have examined the effect of selected therapies on established neuroaxonal dystrophy (NAD) in ileal mesenteric nerves, a rat model of diabetic autonomic neuropathy. Groups of 3-mo-old rats were made diabetic with streptozocin (STZ-D) and allowed to survive untreated for 5 mo, at which time they were begun on sorbinil, dietary myo-inositol, and daily insulin therapies or left untreated for an additional 2 or 4 mo. Ultrastructural evidence of NAD was demonstrated in ileal mesenteric nerves of rats with untreated 5-mo STZ-D and increased with the duration of diabetes. No lesions were demonstrated in control rats of any age. myo-inositol or sorbinil administration failed to alter the severity of diabetes as measured by its metabolic indices. Institution of sorbinil or insulin treatment at 5 mo of diabetes prevented the increase in, but did not normalize, NAD at 7 or 9 mo. Dietary myo-inositol failed to significantly reverse established NAD or prevent its initial development. Morphometric examination of ileal mesenteric nerves demonstrated a decrease in the number of axons comprising each diabetic Schwann cell unit, suggestive of chronic cycles of axonal degeneration and regeneration. This parameter, clearly abnormal by 5 mo of diabetes, was not normalized by 2 or 4 mo of insulin, sorbinil, or myo-inositol treatment. These observations indicate that treatment with insulin or an aldose reductase inhibitor, but not myo-inositol, initiated after the development of structural axonopathy significantly inhibited progression of NAD for the duration of treatment.