Background Pleural effusions (PE) complicate cirrhosis in 5% of patients. Identification of cause and related complications is imperative. Unlike refractory ascites, large-scale studies on interventions for refractory PE are limited. Methods Consecutive hospitalized cirrhotics having PE were retrospectively analysed. None had liver transplantation (LT) within 6-month follow-up. We determined safety, efficacy and mortality predictors for PE managed with standard medical treatment (SMT), thoracentesis, catheter drainage and TIPS. Results Of 1149 cirrhotics with PE (mean Child-Pugh 10.6 +/- 1.8 and MELD 21.2 +/- 7.4), 82.6% had hepatic hydrothorax (HH) and 12.3% were suspected tubercular PE (TBPE). Despite comparable HVPG and MELD scores, patients with HH developed more AKI, encephalopathy and septic shock (all P < .01) on follow-up. Among HH, 73.5% were symptomatic, 53.2% isolated right-sided PE and 21.3% had SBE. Presence of SBP [Odd's ratio, OR: 4.5] and catheter drainage [OR: 2.1] were independent predictors for SBE. In 70.3% of admissions, HH responded to SMT alone, 12.9% required thoracentesis and 11.5% underwent catheter drainage. Fifty-one patients were selected for TIPS [lower mean CTP 9.9 +/- 1.6 and MELD score 18.7 +/- 5.4]. Despite reduction in pressure gradient from 23.1 +/- 3.8 mm Hg to 7.2 +/- 2.5 mm Hg, 25 patients had partial response, 10 had complete HH resolution. Major post-TIPS complications were portosystemic encephalopathy (eight patients, six resolved) and ischaemic hepatitis (four patients, two resolved). Overall, 35.9% patients with HH had 6-month mortality and independent predictors were MELD > 25, SBP and septic shock. Conclusion Refractory PE in cirrhosis requiring interventions including TIPS has poor outcome. The role of haemodynamics in predicting post-TIPS response and complications is limited. Early referral for LT is imperative.