There is renewed interest in the diagnosis of chronic obstructive pulmonary disease (COPD) within primary care. Primary care physicians have difficulty distinguishing asthma from COPD. We tested the feasibility of using spirometry and if appropriate, reversibility testing, to identify patients with COPD on asthma registers in primary care. We carried out a cross-sectional study in three inner-city group practices in east London. Three hundred and twenty-eight patients aged 50 years and over on practice asthma registers were invited to attend for spirometry and, if appropriate, a trial of oral corticosteroids. The main outcome measures were: feasibility of carrying out spirometry; lung function; severity of COPD; prior diagnosis of COPD; response to a corticosteroid trial; quality of life. One hundred and sixty-eight of 328 (51%) patients attended for spirometry. According to British Thoracic Society criteria, 58 (34%) patients had normal spirometry at the time of assessment; 40 (24%) had active asthma and 57 (34%) had COPD. Thirteen patients (8%) were unable to perform spirometry. Of 57 patients with COPD 30 (53%) had mild, 15 (26%) had moderate and 12 (21%) had severe disease. Twenty-three of 57 (40%) patients with COPD on spirometry had this diagnosis recorded prior to the study. New diagnoses of COPD were more likely in those with mild or moderate disease (P<0.05). Twenty-three of 57 (40%) patients with COPD completed a corticosteroid trial: one showed significant reversibility of lung function. Spirometry was feasible and helped identify patients with COPD on asthma registers in these inner-city practices. Patients aged 50 years and over on asthma registers had a wide spectrum of lung function with considerable diagnostic misclassification. Some patients with normal lung function when tested may have had well controlled asthma. New diagnoses of COPD were mainly in those with mild or moderate disease.