Short-term carvedilol administration is more powerful than propranolol in decreasing hepatic venous pressure gradient (HVPG) in cirrhotic patients, but induces arterial hypotension that may prevent its long-term use in portal hypertensive patients. This study compared the HVPG reduction and safety of long-term carvedilol and propranolol. Fifty-one cirrhotic patients were randomly assigned to receive carvedilol (n = 26) and propranolol (n = 25). Hemodynamic measurements and renal function were assessed at baseline and after 11.1 ± 4.1 weeks. Carvedilol caused a greater decrease in HVPG than popranolol (− 19 ± 2% vs. − 12 ± 2%; P < 0.001). The proportion of patients achieving an HVPG reduction ⩾20% or ⩽ 20 mm Hg was greater after carvedilol (54% vs. 23%; P < 0.05). Carvedilol, but not propranolol caused a significant decrease in mean arterial pressure (MAP) (– 11 ± 1% vs. – 5 ± 3%; P = 0.05) and a significant increase in plasma volume (PV) and body weight (11 ± 5% and 2 ± 1%, respectively; P < 0.05). Glomerular filtration rate (GFR) was unchanged with either drug, but the dose of diuretics was increased more frequently after carvedilol (27% vs. 8% P = 0.07). Adverse events requiring discontinuation of treatment occurred in 2 patients receiving carvedilol and in 3 receiving propranolol. In conclusion, carvedilol has a greater portal hypotensive effect than propranolol in patients with cirrhosis. However, its clinical applicability may be limited by its systemic hypotensive effects. Further trials are needed to confirm the therapeutic potential of carvedilol. (HEPATOLOGY2002;36:1367–1373).