Results from FULFIL have shown statistically significant improvements in lung function and health-related quality of life, and a reduction in exacerbation rates with once-daily fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI) 100 µg/62.5 µg/25 µg administered using a single ELLIPTA ® inhaler compared with twice-daily budesonide/formoterol (BUD/FOR) 400 µg/12 µg using the Turbuhaler ® in patients with symptomatic COPD at risk of exacerbations. The safety profile of FF/UMEC/VI reflected that of the components (Lipson, et al. Am J Respir Crit Care Med. 2017). Herein we present post-hoc subgroup analyses of exacerbation rates by prior COPD medication class, disease severity and exacerbation history during FULFIL. In the intent-to-treat (ITT; 24 weeks) population, the mean annual exacerbation rate, FF/UMEC/VI versus BUD/FOR ratios and annual exacerbation rates reductions were calculated for subgroups: by prior COPD medication class, inhaled corticosteroid (ICS) +long acting beta agonists (LABA); BUD/FOR; ICS +LABA + long-acting muscarinic antagonists (LAMA); LAMA; tiotropium; LAMA +LABA; by disease severity, forced expiratory volume in 1 s (FEV 1 ) 1 1 ≥50–≤80%,≥2 moderate or ≥1 severe exacerbations; and by exacerbation history, 0/1 moderate exacerbations;≥2 moderate exacerbations;≥1 severe exacerbation. Up to Week 24 in the ITT population, FF/UMEC/VI versus BUD/FOR improved the mean annual exacerbation rate (range, 63%–24%) in all prior medication subgroups, except LAMA +LABA (annual exacerbation rate reduction, −44%) and improved mean annual exacerbation rates in all disease severity (range, 45%–27%) and exacerbation prior history (range, 57%–27%) subgroups (Table). Statistical significance of the FF/UMEC/VI:BUD/FOR ratio was observed for the subgroups: prior medication class ICS +LABA (0.37; 95% confidence interval [CI] 0.20–0.71; p=0.003) and ICS +LAMA + LABA (0.53; 95% CI 0.33–0.87; p=0.012); disease severity FEV 1 Funding GSK (NCT02345161; CTT116853) Please refer to page A260 for declarations of interest in relation to abstract P272.