1. We have employed a thermal clearance probe to study fingertip blood flow, which comprises predominantly arteriovenous anastomotic flow.
2. In non-diabetic subjects there was a marked variability in resting fingertip blood flow which was of similar magnitude to that in diabetic subjects without autonomic neuropathy.
3. In 20 diabetic subjects with increasing severity of autonomic neuropathy there was a reduction of spontaneous variability in flow (r= 0.59, P < 0.001), which was not related to differences in basal flow, in skin thickness or in age.
4. There was no relationship between maximal flow and severity of autonomic neuropathy (r= 0.08), although basal flow was decreased in subjects with moderate, but not those with severe, autonomic neuropathy (P < 0.01 and P > 0.4 compared with controls).
5. Our failure to find an increase in basal fingertip flow in autonomic neuropathy could imply that spontaneous variability of anastomic flow is lost before increased arteriovenous shunting becomes apparent, but it could result from the indirect heating employed in the study causing vasodilatation in subjects without neuropathy.