Eosinophilic bronchitis is a common cause of chronic cough, characterized by sputum eosinophilia similar to that seen in asthma, but unlike asthma the patients have no objective evidence of variable airflow obstruction or airway hyperresponsiveness. The reason for the different functional associations is unclear. The authors have tested the hypothesis that in eosinophilic bronchitis the inflammation is mainly localized in the upper airway.In an open study the authors measured the lower (provocative concentration causing a 20% fall in forced expiratory volume in one second (PC20)) and upper (PC25 MIF50) airway responsiveness to histamine, lower and upper airway inflammation using induced sputum and nasal lavage, in 11 patients with eosinophilic bronchitis. The authors assessed changes in these measures and in cough reflex sensitivity to capsaicin and cough severity after 400 mg of inhaled budesonide for 4 weeks.A nasal eosinophilia was present in only three patients with one having upper airway hyperresponsiveness. Following treatment with inhaled corticosteroids the geometric mean sputum eosinophil count decreased from 12.8% to 2.9% (mean difference 4.4-fold, 95% confidence interval (CI) 2.14±10.02), the mean SEM cough visual analogue score on a 100 mm scale decreased from 27.2 6.6 mm to 12.6 5.7 mm (mean difference 14.6, 95% CI 9.1±20.1) and the cough sensitivity assessed as the capsaicin concentration required to cause two coughs (C 2 ) and five coughs (C 5 ) improved (C 2 mean difference 0.75 doubling concentrations, 95% CI 0.36±1.1; C 5 mean difference 1.3 doubling concentration, 95% CI 0.6±2.1). There was a significant positive correlation between the fold change in sputum eosinophil count and doubling dose change in C 5 after inhaled budesonide (r=0.61).It is concluded that upper airway inflammation is not prominent in eosinophilic bronchitis and that inhaled budesonide improves the sputum eosinophilia, cough severity and sensitivity suggesting a causal link between the inflammation and cough. Eur Respir J 2000; 15: 682±686.