Benefits of triple therapy with a long-acting muscarinic antagonist (LAMA), added to inhaled corticosteroid (ICS)/long-acting β2-agonist (LABA), have been demonstrated. Limited data assessing the efficacy of the LAMA umeclidinium (UMEC) added to ICS/LABA are available. The aim of this study is to evaluate the efficacy and safety of UMEC added to ICS/LABAs in patients with moderate-to-very-severe COPD. This is a multicentre, randomised, double-blind, parallel-group study. Patients were symptomatic (modified Medical Research Council Dyspnoea Scale score ⩾2), despite receiving ICS/LABA (fluticasone propionate/salmeterol (FP/SAL, branded) 500/50 mcg, budesonide/formoterol (BD/FOR, branded) 200/6 mcg or 400/12 mcg, or other ICS/LABAs) ⩾30 days before the run-in (7±2 days). Patients were randomised 1:1 to once-daily UMEC 62.5 mcg or placebo (PBO), added to twice-daily open-label ICS/LABA for 12 weeks. Primary end point was trough forced expiratory volume in 1 s (FEV1) at Day 85; secondary end point was weighted mean (WM) 0-6 h FEV1 at Day 84; other end points included COPD Assessment Test (CAT) score and Transition Dyspnoea Index (TDI) score. Adverse events (AEs) were investigated. In the UMEC+ICS/LABA and PBO+ICS/LABA groups, 119 and 117 patients were randomised, respectively. Patients received FP/SAL (40%), BD/FOR (43%) and other ICS/LABAs (17%). UMEC+ICS/LABA resulted in significant improvements in trough FEV1 (Day 85) and in WM 0-6 h FEV1 (Day 84) versus PBO+ICS/LABA (difference: 123 and 148 ml, respectively, both P<0.001). Change from baseline for UMEC+ICS/LABA versus PBO+ICS/LABA was significantly different for CAT score at Day 84 (-1.31, P<0.05), but not for TDI score (0.40, P=0.152). AE incidence was similar with UMEC+ICS/LABA (38%) and PBO+ICS/LABA (42%). UMEC+ICS/LABA improved lung function and CAT score in patients with symptomatic COPD versus PBO+ICS/LABA (ClinicalTrials.gov NCT02257372).