||BACKGROUND: Recommendations on interpreting tests of bronchodilator responsiveness (BDR) are conflicting. We investigated the dependence of BDR criteria on sex, age, height, ethnicity, and severity of respiratory impairment. METHODS: BDR test data were available from clinical patients in the Netherlands, New Zealand, and the United States (n = 15,278; female subjects, 51.7%) and from surveys in Canada, Norway, and five Latin-American countries (n = 16,250; female subjects, 54.7%). BDR calculated according to FEV1, FVC, and FEV1/FVC was expressed as absolute change, a percentage of the baseline level (% baseline), a percentage of the predicted value (% predicted), and z score. RESULTS: Change (Delta) in FEV1 and FVC, in milliliters, was unrelated to the baseline value but was biased toward age, height, sex, and level of airways obstruction; DeltaFEV1 was significantly lower in African Americans. In 1,106 subjects with low FEV1 (200-1,621 mL) the FEV1 increased by 12% to 44.7% relative to baseline but < 200 mL. Expressing BDR as a percentage of the predicted value or as a z score attenuated the bias and made the 200-mL criterion redundant, but reduced positive responses by half. DeltaFEV1 % baseline increased with the level of airflow obstruction but decreased with severe obstruction when expressed as z scores or % predicted; DeltaFVC, however expressed, increased with the level of airflow obstruction. CONCLUSIONS: Expressing FEV1 responsiveness as % baseline spuriously suggests that responsiveness increases with the severity of respiratory impairment. Expressing change in FEV1 or FVC as % predicted or as z scores eliminates this artifact and renders the required 200-mL minimum increase redundant. In severe airways obstruction DeltaFVC should be critically evaluated as an index of clinically important relief of hyperinflation, with implications for bronchodilator drug trials.