Abstract Background Short-acting β2-agonist (SABA) bronchodilators help alleviate symptoms in chronic obstructive pulmonary disease (COPD) and may be a useful marker of symptom severity. This analysis investigated whether SABA use impacts treatment differences between maintenance dual- and mono-bronchodilators in patients with COPD. Methods The Early MAXimisation of bronchodilation for improving COPD stability (EMAX) trial randomised symptomatic patients with low exacerbation risk not receiving inhaled corticosteroids 1:1:1 to once-daily umeclidinium/vilanterol 62.5/25 μg, once-daily umeclidinium 62.5 μg or twice-daily salmeterol 50 μg for 24 weeks. Pre-specified subgroup analyses stratified patients by median baseline SABA use (low, < 1.5 puffs/day; high, ≥1.5 puffs/day) to examine change from baseline in trough forced expiratory volume in 1 s (FEV1), change in symptoms (Transition Dyspnoea Index [TDI], Evaluating Respiratory Symptoms-COPD [E-RS]), daily SABA use and exacerbation risk. A post hoc analysis used fractional polynomial modelling with continuous transformations of baseline SABA use covariates. Results At baseline, patients in the high SABA use subgroup (mean: 3.91 puffs/day, n = 1212) had more severe airflow limitation, were more symptomatic and had worse health status versus patients in the low SABA use subgroup (0.39 puffs/day, n = 1206). Patients treated with umeclidinium/vilanterol versus umeclidinium demonstrated statistically significant improvements in trough FEV1 at Week 24 in both SABA subgroups (59–74 mL; p < 0.001); however, only low SABA users demonstrated significant improvements in TDI (high: 0.27 [p = 0.241]; low: 0.49 [p = 0.025]) and E-RS (high: 0.48 [p = 0.138]; low: 0.60 [p = 0.034]) scores. By contrast, significant reductions in mean SABA puffs/day with umeclidinium/vilanterol versus umeclidinium were observed only in high SABA users (high: − 0.56 [p < 0.001]; low: − 0.10 [p = 0.132]). Similar findings were observed when comparing umeclidinium/vilanterol and salmeterol. ...