Abstract The prevalence of COPD is high, and most cases remain undiagnosed. In contrast, some patients labeled and treated as COPD do not have spirometric confirmation. Our objective was to determine the prevalence of COPD among smokers aged 45 years or older and investigate the accuracy of diagnosis of COPD in primary care. A population-based, epidemiological study was conducted in a primary care centre among subjects older than 45 years with a history of smoking. The participants underwent a clinical questionnaire and spirometry with bronchodilator test. Additionally, participants with newly diagnosed COPD, defined as postbronchodilator FEV1/FVC<0.7, underwent 4-week treatment with formoterol and budesonide to rule out reversible airflow obstruction. A total of 1,738 individuals (84.4% male) with a mean age of 59.9 years were included. The prevalence of COPD was 24.3% (95%, CI 22.3-26.4), with an overall underdiagnosis of 56.7%. Patients with COPD were older, more frequently male, with a lower body mass index, a longer history of smoking, lower educational level, previous occupational exposure, and more cardiovascular co-morbidity (all p < 0.001). After 4 weeks of treatment, 16% of initially obstructed patients had normal spirometry; in addition, 15.6% of individuals with a diagnosis of COPD did not have airflow obstruction. One out of four smokers 45 years or older presenting in primary care have airflow obstruction, mostly undiagnosed. However, among those with an initial diagnosis of COPD up to 16% will normalise spirometry after 4 weeks of treatment. There is also a significant number of individuals misdiagnosed with COPD.