OBJECTIVE Vasopressin reduces portal pressure which may be due to decreased portal inflow. However, it remains unclear whether vasopressin is able to selectively reduce esophageal varices blood flow. The aim of this study was to address this question. METHODS Fifteen patients with cirrhosis and esophageal varices were included in this prospective study. Portal vein and superior mesenteric artery flow velocity were measured with a percutaneous echo-Doppler. Esophageal varices flow velocity was measured using a transesophageal echo-Doppler technique. Mean arterial pressure and heart rate were also recorded. These measurements were performed at baseline condition and 15 min after observer blind drug administration. In this study, two groups, six patients receiving placebo and nine patients receiving 0.3 U/min of vasopressin, were randomized according to the coded number. RESULTS Placebo administration had no effect on systemic and splanchnic circulation. In contrast, vasopressin administration increased mean arterial pressure (p < 0.05) associated with a bradycardia (p < 0.01). In splanchnic circulation, vasopressin decreased portal vein (-32 +/- 3%, p < 0.01), superior mesenteric artery (-30 +/- 2%, p < 0.01), and esophageal varices flow velocity (-48 +/- 5%, p < 0.01). When the magnitude of these reductions was compared, ANOVA showed a significant difference (p < 0.01). Furthermore, the reduction in esophageal varices flow velocity was significantly higher than that in portal vein flow velocity (p < 0.01) and that in superior mesenteric artery flow velocity (p < 0.01). CONCLUSIONS These data support the view that vasopressin is able to selectively reduce esophageal varices blood flow. This effect, in addition to its well-established portal pressure reducing action, may play a role in its therapeutic efficacy in the treatment of variceal bleeding.