Background Despite widespread use of therapies such as inhaled corticosteroids (ICSs), people with chronic obstructive pulmonary disease (COPD) continue to suffer, have reduced life expectancy and utilise considerable NHS resources. Laboratory investigations have demonstrated that at low plasma concentrations (1–5 mg/l) theophylline markedly enhances the anti-inflammatory effects of corticosteroids in COPD. Objective To determine the clinical effectiveness and cost-effectiveness of adding low-dose theophylline to a drug regimen containing ICSs in people with COPD at high risk of exacerbation. Design A multicentre, pragmatic, double-blind, randomised, placebo-controlled clinical trial. Setting The trial was conducted in 121 UK primary and secondary care sites. Participants People with COPD [i.e. who have a forced expiratory volume in 1 second (FEV₁)/forced vital capacity (FVC) of < 0.7] currently on a drug regimen including ICSs with a history of two or more exacerbations treated with antibiotics and/or oral corticosteroids (OCSs) in the previous year. Interventions Participants were randomised (1 : 1) to receive either low-dose theophylline or placebo for 1 year. The dose of theophylline (200 mg once or twice a day) was determined by ideal body weight and smoking status. Primary outcome The number of participant-reported exacerbations in the 1-year treatment period that were treated with antibiotics and/or OCSs. Results A total of 1578 people were randomised (60% from primary care): 791 to theophylline and 787 to placebo. There were 11 post-randomisation exclusions. Trial medication was prescribed to 1567 participants: 788 in the theophylline arm and 779 in the placebo arm. Participants in the trial arms were well balanced in terms of characteristics. The mean age was 68.4 [standard deviation (SD) 8.4] years, 54% were male, 32% smoked and mean FEV₁ was 51.7% (SD 20.0%) predicted. Primary outcome data were available for 98% of participants: 772 in the theophylline arm and 764 in the placebo arm. There were 1489 person-years of follow-up data. The mean number of exacerbations was 2.24 (SD 1.99) for participants allocated to theophylline and 2.23 (SD 1.97) for participants allocated to placebo [adjusted incidence rate ratio (IRR) 0.99, 95% confidence interval (CI) 0.91 to 1.08]. Low-dose theophylline had no significant effects on lung function (i.e. FEV₁), incidence of pneumonia, mortality, breathlessness or measures of quality of life or disease impact. Hospital admissions due to COPD exacerbation were less frequent with low-dose theophylline (adjusted IRR 0.72, 95% CI 0.55 to 0.94). However, 39 of the 51 excess hospital admissions in the placebo group were accounted for by 10 participants having three or more exacerbations. There were no differences in the reporting of theophylline side effects between the theophylline and placebo arms. Limitations A higher than expected percentage of participants (26%) ceased trial medication; this was balanced between the theophylline and placebo arms and mitigated by over-recruitment (n = 154 additional participants were recruited) and the high rate of follow-up. The limitation of not using documented exacerbations is addressed by evidence that patient recall is highly reliable and the results of a small within-trial validation study. Conclusion For people with COPD at high risk of exacerbation, the addition of low-dose oral theophylline to a drug regimen that includes ICSs confers no overall clinical or health economic benefit. This result was evident from the intention-to-treat and per-protocol analyses. Future work To promote consideration of the findings of this trial in national and international COPD guidelines. Trial registration Current Controlled Trials ISRCTN27066620. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 23, No. 37. See the NIHR Journals Library website for further project information.