OBJECTIVES: Chronic obstructive pulmonary disease (COPD) is one of the leading causes of morbidity and mortality worldwide. Long-acting bronchodilators (LABs) are the mainstay of pharmacological maintenance therapy for COPD. However, the possibility that use of LABs may lead to risk of cardiovascular events remains debated. The objective is to provide a critical review of methodologies employed to evaluate risk of cardiovascular events in patients with COPD using LABs. METHODS: A targeted search was conducted in PubMed to identify all original published research reporting adverse cardiovascular events from clinical trials and observational studies. The search was limited to English language, but not restricted by publication date. Selected abstracts were reviewed for study population, treatment, follow-up duration, study design, and cardiovascular events. RESULTS: The search returned 131 citations, of which 19 reported results from original research. Among them, 10 studies (53%) were clinical trials and 9 (47%) were observational studies. For the clinical trials, sample sizes ranged from 204 to 6,184 patients; follow-up duration was 14 days to 4 years. For the observational studies, 6 (67%) were nested case-control studies, 2 (22%) cohort studies, 1 (11%) self-controlled case series study; sample sizes ranged from 1,043 to 352,631 patients; follow-up duration was 52 weeks to 13 years. The most frequently reported cardiovascular events included arrhythmias, stroke, angina, myocardial infarction and cardiovascular death. While none of the clinical trials showed a statistically significant association between the use of LABs and increased risk of cardiovascular events, 5 observational studies did.  CONCLUSIONS: The controversy continues regarding the use of LABs and their cardiovascular safety. Short follow-up duration and exclusion of patients with previous history of cardiovascular events are the main limitations to clinical trials, while observational studies may be limited by residual confounding by disease severity and immeasurable time bias. A well-designed observational study is warranted.