To image esophageal and gastric varices qualitatively and to measure esophageal varices quantitatively, 29 patients with portal hypertension underwent transnasal esophageal and gastric ultrasonography (US). Twenty-three patients underwent standard endoscopy. Endoluminal US demonstrated the varices as anechoic areas with communications in the submucosal, periesophageal, and perigastric regions, as well as ascites visualized through the gastric wall. Interobserver variation between two investigators for measurements of the largest esophageal varix in each patient was r value of .99 for diameter, r value of .99 for cross-sectional surface area, and r value of .98 for circumference. Endoscopy for evaluation of gastric varices had a sensitivity of 48% and a specificity of 50% and for evaluation of esophageal varices had a sensitivity of 94% and a specificity of 17%. Periesophageal and perigastric varices could not be visualized at endoscopy. Transnasal esophageal US, a new imaging technique with which to detect and measure esophageal varices, is a more sensitive modality than endoscopy for the detection of gastric varices.