This study evaluated the clinical application of a pressure-sensitive gauge that allows the noninvasive measurement of the pressure of esophageal varices at endoscopy. The study was performed in 70 patients with cirrhosis and portal hypertension. Among them, 47 had bled from the varices and 23 had varices but had not bled. In addition to measurements of variceal pressure, the size of the varices was estimated semiquantitatively at endoscopy. This allowed an estimate of the tension on the wall of the varices as the product of the transmural pressure and the estimated radius of the varices. Most patients had a standard hemodynamic evaluation of portal hypertension, with measurements of wedged and free hepatic venous pressures, and of azygos blood flow. These were performed within 24 h of the variceal pressure measurements. Variceal pressure was significantly higher in bleeders than in nonbleeders (15.7 +/- 2.8 vs. 12.1 +/- 2.6 mmHg, p less than 0.001) in spite of a similar portal pressure in both groups (20.1 +/- 5.1 vs. 20.4 +/- 7.6 mmHg, NS). More than 60% of the bleeders, but only 22% of the nonbleeders had a variceal pressure greater than or equal to 15 mmHg (p less than 0.005). Among nonbleeders, variceal pressure was higher in patients with large varices (13.9 +/- 2 mmHg, n = 9) than in those with small varices (10.9 +/- 2.4 mmHg, n = 14) (p less than 0.01). Estimates of variceal wall tension further exaggerated the differences between bleeders and nonbleeders (66.1 +/- 22.6 vs. 32.0 +/- 19.8 mmHg.mm, p less than 0.001). More than 50% of bleeders, but just 9% of nonbleeders had an estimated variceal tension greater than 50 mmHg.mm (p less than 0.001). Our findings support the role of an increased variceal pressure in the pathogenesis of variceal hemorrhage, and suggest that this noninvasive technique can be valuable in assessing the risk of variceal hemorrhage in patients with portal hypertension.