Background: We compared the efficacy and safety of indacaterol and tiotropium in patients with severe chronic obstructive pulmonary disease (COPD) and a history of at least one moderate to severe exacerbation in the previous 12 months. Methods: In this multicentre, randomised, blinded, double-dummy, parallel group study, we enrolled patients aged 40 years or older with severe COPD and at least one exacerbation within the previous year. We used a computer-generated sequence to randomly allocate patients (1:1; stratified by baseline inhaled corticosteroid use, with the balance of treatments maintained at country level) to receive either indacaterol (150 μg) or tiotropium (18 μg) once-daily for 52 weeks. Our primary and key secondary objectives were to investigate whether indacaterol was non-inferior to tiotropium for trough forced expiratory volume in 1 s (FEV1) at week 12 (primary endpoint), and for rate of exacerbations at week 52 (secondary endpoint). Analysis populations for the primary and key secondary endpoints were per-protocol sets. The safety set included all patients who received at least one dose of study drug. This study is registered with ClinicalTrials.gov, number NCT00845728. Findings: Between March 16, 2009, and July 5, 2012, we enrolled and randomly allocated 3444 patients: 1723 to indacaterol and 1721 to tiotropium. At week 12, the estimated least squares mean trough FEV1 difference between the groups was -0·011 L (least squares mean with indacaterol [n=1450] 1·134 L [SE 0·008] vs tiotropium [n=1467] 1·145 L [0·008]; one-sided 97·5% CI lower limit -0·026 L; p<0·0001). The lower limit of the 97·5% CI was above the prespecified non-inferiority margin of -0·055 L, suggesting that indacaterol was non-inferior to tiotropium. Indacaterol did not show non-inferiority in terms of annualised exacerbation rates: 0·79 (indacaterol, n=1529) versus 0·61 (tiotropium, n=1543); ratio 1·29 (one-sided 97·5% CI upper limit 1·44). In the safety set, we recorded no between-group difference in the number of patients who had adverse events (indacaterol 1119 [65%] of 1721 patients vs tiotropium 1065 [62%] of 1718 patients) or serious adverse events (indacaterol, 263 [15%] of 1721 patients vs tiotropium, 255 [15%] of 1718 patients). Respiratory disorders, particularly worsening of COPD, were the most common adverse events (COPD: indacaterol, 747 [43%] of 1721 patients and tiotropium, 665 [39%] of 1718 patients) and serious adverse events (COPD: indacaterol, 147 [9%] of 1721 patients and tiotropium, 121 [7%] of 1718 patients). Interpretation: Indacaterol and tiotropium provided clinically relevant improvements in lung function with comparable safety profiles. Tiotropium afforded greater protection from exacerbations, although the absolute number of events was small and the difference between treatments is of uncertain clinical importance. The present data offer evidence consistent with current guidelines. Funding: Novartis Pharma AG.