BACKGROUND: Compared to inhalation and local anesthetics, little is known about the mechanisms of vascular effects of intravenous anesthetics. So we studied the effects of thiopental sodium, midazolam, propofol and ketamine on the endothelial nitric oxide-cGMP pathway and also on the membrane cyclooxygenase pathway. METHODS: After isolating ring strips of rat thoracic aorta, we measured the relaxation ED50 values of the four intravenous anesthetics from the maximally contracted using phenylephrine 10(-5)M. Then using L-NAME and methylene blue, we studied the effects of the drugs upon the NO-cGMP system. In addition, another pathway of vasodilation through membrane prostaglandin metabolism was examined using the membrane cyclooxygenase inhibitor, indomethacine. RESULTS: The following results were obtained. 1. Thiopental sodium (10(-5)M) did not have any effect on the PE induced contractions of aortic rings but midazolam (10(-6)M), propofol (10(-4)M) and ketamine (10(-3)M) significantly (P < 0.05) inhibited the PE induced contractions of aortic rings. 2. Midazolam 10(-6)M and propofol 10(-4)M induced relaxation of aortic rings were recovered with L-NAME pretreatment but ketamine induced relaxation was not recovered with L-NAME. 3. Midazolam 10(-6)M induced relaxation was not recovered with methylene blue pretreatment, but propofol 10(-4)M induced relaxation was recovered with methylene blue. 4. Indomethacine pretreatment induced further relaxation of midazolam or propofol induced relaxation of aortic rings. CONCLUSIONS: Midazolam, propofol and ketamine, but not thiopental sodium, relax rat thoracic aortic rings, and these relaxation effects of midazolam and propofol are endothelium dependent. Cyclooxygenase inhibition is related at least in part to midazolam or propofol induced relaxation, and guanylate cyclase to propofol induced relaxation.