We report the results of adapting medical therapy to the monitoring of hemodynamic response in the prevention of a first variceal bleeding or rebleeding in patients with cirrhosis. Hepatic venous pressure gradient (HVPG) was measured before and after propranolol was initiated. The patients were considered responders if HVPG decreased below 12 mm Hg or at least 20% as compared with baseline value. If patients were not responders, isosorbide-5 mononitrate (I-5MN) was added, and a third hemodynamic study was performed. Thereafter, the patients were followed for a mean of 28 months. Thirty-four consecutive patients were treated to prevent a first bleeding episode in 20 patients and a rebleeding in 14 patients. HVPG value was initially 19.8 +/- 4.6 mm Hg and decreased to 17.6 +/- 5.7 mm Hg (P <.05) after propranolol alone. Thirteen patients (38%) were responders to propranolol. I-5MN improved hemodynamic response in 7 cases. Among these 20 (59%) hemodynamic responders, only 2 (10%) experienced variceal bleeding, as compared with 9 of 14 (64%) nonresponders (P <.05). Using multivariate analysis, only hemodynamic response was found to have an independent predictive value for the risk of variceal bleeding. In conclusion, hemodynamic response to drug therapy identifies patients who are efficiently protected from variceal bleeding as well as nonresponders in whom an alternative treatment should be considered.