Although it is generally accepted that isoflurane can cause cerebral vasodilation, the sensitivity of the cerebral vessels to this anesthetic agent remains controversial. Furthermore, the mechanism by which isoflurane produces its direct effects on the cerebral vasculature remains unknown. The purpose of this study was to determine if isoflurane-induced relaxation of canine middle cerebral arteries is dose-dependent and/or endothelium-dependent. In an additional series of experiments, isoflurane-induced relaxation was studied in the presence of indomethacin to inhibit prostacyclin release, and endothelium-independent relaxation was examined with sodium nitroprusside. The response to isoflurane was examined in middle cerebral arteries prior to and following pretreatment with 300 microM NG-monomethyl-L-arginine (LnMMA), an inhibitor of endothelium-dependent vasodilation. Vascular rings (2.5 mm in length and 600-800 microns in diameter) were suspended in tissue baths and isometric tension recorded. The rings were constricted with either 0.2 microM 5-hydroxytryptamine or 5 microM prostaglandin F2 alpha and subsequently exposed to increasing concentrations of isoflurane (0.65-4.9%). In separate experiments the procedure was repeated in vessels with and without endothelium. Isoflurane produced a dose-dependent relaxation in all vessels. This relaxation was not inhibited by LnMMA and was unaffected by the absence of endothelium. The isoflurane response was independent of cyclooxygenase inhibition. These results demonstrate that isoflurane-induced relaxation of canine middle cerebral arteries: 1) is dose-dependent; 2) is not mediated by modulation of endothelium-derived relaxing factor or a release of prostacyclin; and 3) is endothelium-independent.