Increased portal pressure during variceal bleeding may have an influence on the treatment failure rate, as well as on short- and long-term survival. However, the usefulness of hepatic hemodynamic measurement during the acute episode has not been prospectively validated, and no information exists about the outcome of hemodynamically defined high-risk patients treated with early portal decompression. Hepatic venous pressure gradient (HVPG) measurement was made within the first 24 hours after admission of 116 consecutive patients with cirrhosis with acute variceal bleeding treated with a single session of sclerotherapy injection during urgent endoscopy. Sixty-four patients had an HVPG less than 20 mm Hg (low-risk [LR] group), and 52 patients had an HVPG greater than or equal to 20 mm Hg (high-risk [HR] group). HR patients were randomly allocated into those receiving transjugular intrahepatic portosystemic shunt (TIPS; HR-TIPS group, n = 26) within the first 24 hours after admission and those not receiving TIPS (HR-non-TIPS group). The HR-non-TIPS group had more treatment failures (50% vs. 12%, P =.0001), transfusional requirements (3.7 +/- 2.7 vs. 2.2 +/- 2.3, P =.002), need for intensive care (16% vs. 3%, P <.05), and worse actuarial probability of survival than the LR group. Early TIPS placement reduced treatment failure (12%, P =.003), in-hospital and 1-year mortality (11% and 31%, respectively; P <.05). In conclusion, increased portal pressure estimated by early HVPG measurement is a main determinant of treatment failure and survival in variceal bleeding, and early TIPS placement reduces treatment failure and mortality in high risk patients defined by hemodynamic criteria.