Hypertension is associated with endothelial dysfunction that is attributable to oxidative stress and a proinflammatory state. Under these conditions, enhanced expression of cyclooxygenase-2 might lead to increased production of vasoconstrictor prostanoids and reactive oxygen species that reduce the bioavailability of endothelium-derived nitric oxide. To investigate the contribution of cyclooxygenase-2 activity to endothelial dysfunction in human hypertension, we evaluated brachial artery vasodilator function by ultrasound in 29 hypertensive patients before and after treatment with the selective cyclooxygenase-2 inhibitor celecoxib or placebo in a randomized, double-blind study. Brachial artery flow-mediated dilation improved from a baseline of 7.9+/-4.5% to 9.9+/-5.1% (P=0.005) 3 hours after the first dose and to 10.1+/-6.1% (P=0.006) after 1 week of treatment with celecoxib. In contrast, placebo treatment had no significant effect on flow-mediated dilation (8.1+/-4.4%, 8.3+/-3.5%, and 8.0+/-3.2%, respectively). Neither treatment altered nitroglycerin-mediated dilation, extent of reactive hyperemia, or baseline arterial diameter. Celecoxib treatment had no significant effect on the urinary concentrations of F2 isoprostane or thromboxane metabolites. However, urinary concentrations of the prostacyclin metabolite 2,3-dinor-6-ketoprostglandin F1alpha were significantly lower after 1 week of celecoxib treatment. Thus, cyclooxygenase-2 products contribute to endothelial dysfunction in hypertension, and treatment with a cyclooxygenase-2 inhibitor could have a beneficial effect in this setting. However, cyclooxygenase-2 inhibition also has an adverse effect on prostacyclin production that could promote thrombosis, and the net clinical consequences of improved endothelial function versus loss of prostacyclin merits further investigation.