We have demonstrated previously, using dog epicardial arteries of different sizes, that isoflurane, like adenosine, is preferentially a small coronary artery dilator, whereas halothane, like nitroglycerin, is a large artery dilator. The present study was designed to compare the direct effects of sevoflurane with those of isoflurane and halothane. Proximal large coronary arteries with an outer diameter (o.d.) of 2.5-3.2 mm and distal small arteries of 0.6-0.9 mm o.d. were isolated from dogs and then cut into vascular rings. They were precontracted with KCl (20 mM), and their relaxant responses to anaesthetics were compared relative to the maximal responses induced by papaverine. Sevoflurane, halothane and isoflurane (1-3 human MAC) induced dose-dependent relaxation of these arteries. The relaxant response to sevoflurane did not differ between large and small arteries. However, the relaxant response of the large arteries to halothane (1.5-2.3%) was greater than that of small arteries (P < 0.01) and the response of small arteries to isoflurane (3.5%) was greater than that of large arteries (P < 0.05). In large arteries, the potency of the relaxant effect at equivalent human MAC could be ranked as halothane > or = sevoflurane > isoflurane, and, in small epicardial arteries as isoflurane > sevoflurane >> halothane. We conclude that, unlike isoflurane, sevoflurane is not a preferential dilator of small coronary arteries.