© 2019 Elsevier Ltd Background: Clinical decompensation of cirrhosis is associated with poor prognosis. Clinically significant portal hypertension (CSPH), defined by a hepatic venous pressure gradient (HVPG) ≥10 mm Hg, is the strongest predictor of decompensation. This study aimed at assessing whether lowering HVPG with β blockers could decrease the risk of decompensation or death in compensated cirrhosis with CSPH. Methods: This study on β blockers to prevent decompensation of cirrhosis with portal hypertension (PREDESCI) was an investigator-initiated, double-blind, randomised controlled trial done in eight hospitals in Spain. We enrolled patients with compensated cirrhosis and CSPH without high-risk varices. All participants had HVPG measurements with assessment of acute HVPG-response to intravenous propranolol. Responders (HVPG-decrease ≥10%) were randomly assigned to propranolol (up to 160 mg twice a day) versus placebo and non-responders to carvedilol (≤25 mg/day) versus placebo. Doses were individually determined during an open-label titration period after which randomisation was done with 1:1 allocation by a centralised web-based system. The primary endpoint was incidence of cirrhosis decompensation (defined as development of ascites, bleeding, or overt encephalopathy) or death. Since death in compensated cirrhosis is usually unrelated to the liver, an intention-to-treat analysis considering deaths unrelated to the liver as competing events was done. This study is registered with ClinicalTrials.gov, number NCT01059396. The trial is now completed. Findings: Between Jan 18, 2010, and July 31, 2013, 631 patients were evaluated and 201 were randomly assigned. 101 patients received placebo and 100 received active treatment (67 propranolol and 33 carvedilol). The primary endpoint occurred in 16 (16%) of 100 patients in the β blockers group versus 27 (27%) of 101 in the placebo group (hazard ratio [HR] 0·51, 95% CI 0·26–0·97, p=0·041). The difference was due to a reduced incidence of ascites (HR=0·44, 95%CI=0·20–0·97, p=0·0297). The overall incidence of adverse events was similar in both groups. Six patients (four in the β blockers group) had severe adverse events. Interpretation: Long-term treatment with β blockers could increase decompensation-free survival in patients with compensated cirrhosis and CSPH, mainly by reducing the incidence of ascites. Funding: Spanish Ministries of Health and Economy.