Objectives To determine whether acute exacerbations of chronic obstructive pulmonary disease (AECOPD) triggered by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), have worse outcomes than AECOPD caused by other infectious agents or non-infective AECOPD (NI-COPD). Design A two-hospital prospective cohort study of adults hospitalised with acute respiratory disease. We compared outcomes with AECOPD and a positive test for SARS-CoV-2 (n = 816), AECOPD triggered by other infections (n = 3038) and NI-COPD (n = 994). We used multivariable modelling to adjust for potential confounders and assessed variation by seasons associated with different SARS-CoV-2 variants. Setting Bristol UK, August 2020–May 2022. Participants Adults (≥18 y) hospitalised with AECOPD. Main outcome measures We determined the risk of positive pressure support, longer hospital admission and mortality following hospitalisation with AECOPD due to non-SARS-CoV-2 infection compared with SARS-CoV-2 AECOPD and NI-COPD. Results Patients with SARS-CoV-2 AECOPD, in comparison to non-SARS-CoV-2 infective AECOPD or NI-COPD, more frequently required positive pressure support (18.5% and 7.5% vs. 11.7%, respectively), longer hospital stays (median [interquartile range, IQR]: 7 [3–15] and 5 [2–10] vs. 4 [2–9] days, respectively) and had higher 30-day mortality (16.9% and 11.1% vs. 5.9%, respectively) (all p < 0.001). In adjusted analyses, SARS-CoV-2 AECOPD was associated with a 55% (95% confidence interval [95% CI]: 24–93), 26% (95% CI: 15–37) and 35% (95% CI: 10–65) increase in the risk of positive pressure support, hospitalisation length and 30-day mortality, respectively, relative to non-SARS-CoV-2 infective AECOPD. The difference in risk remained similar during periods of wild-type, Alpha and Delta SARS-CoV-2 strain dominance, but diminished during Omicron dominance. Conclusions SARS-CoV-2-related AECOPD had worse patient outcomes compared with non-SARS-CoV-2 AECOPD or NI-AECOPD, although the difference in risks was less pronounced during Omicron dominance.