Furchgott and Zawadski have shown that acetylcholine (ACh) does not act directly on the smooth muscle of blood vessel walls, but rather via receptors on the endothelial cells lining the lumen, to release an endothelium-derived relaxing factor (EDRF). As it is very unlikely that neurotransmitter released from the periarterial nerves, which are confined to the adventitial-medial border, diffuses all the way through the medial muscle coat before acting on endothelial cells to release EDRF to produce vasodilatation, this discovery has been regarded as an indication of a pathophysiological mechanism, rather than a physiological one (see refs 2, 3). ACh is rapidly degraded in the blood by acetylcholinesterase, so that ACh must be released locally to be effective on endothelial cells. Here we demonstrate the immunocytochemical localization of choline acetyltransferase in endothelial cells of small brain vessels, which is consistent with the view that the ACh originates from endothelial cells that can synthesize and store it. We suggest that release of ACh following damage to endothelial cells during ischaemia contributes to a pathophysiological mechanism of vasodilation which protects that segment of vessel from further damage as well as brain cells from hypoxia.