Severe microvascular disease exists at the stage of clinical diabetic neuropathy. A non-invasive test that will identify those diabetic subjects who will eventually develop neuropathy is essential for early intervention. Sural sensory conduction velocity was recorded (x3) in 12 non-neuropathic diabetic subjects, 15 diabetic subjects with established neuropathy and 16 age-matched normal control subjects, before and after exercise to 80% age/sex predicted maximum heart rate. Fixed sural electrodes were used. Subcutaneous temperature was recorded by a needle thermocouple placed near the sural nerve. Sural sensory conduction velocity increased significantly after exercise in normal subjects (p less than 0.01, mean increase 5.07 m/s) and non-neuropathic diabetic subjects (p less than 0.02, mean increase 3.99 m/s) but not in neuropathic subjects (mean increase 0.99 m/s). Subcutaneous temperature rose significantly in normal subjects (p less than 0.01, mean increase 2.07 degrees C) and non-neuropathic diabetic subjects (p less than 0.001, mean increase 2.52 degrees C) but not in neuropathic subjects (mean increase 0.15 degree C). However, sural sensory conduction velocity increased by 1.2 m.s-1.degree C-1 following direct warming of the limb in six neuropathic subjects which was comparable to that of normal and non-neuropathic subjects (1.49 and 1.48 m.s-1. degree C-1). The impairment of exercise conduction increment in diabetic neuropathy suggests impaired nerve blood flow in diabetic neuropathy.