Bronchial inflammation in mild asthma has been investigated using bronchoscopical techniques. The safety of bronchoscopy in patients with more severe asthma has been questioned. We have used the non-invasive technique of hypertonic saline (HS) inhalation to induced sputum samples for cellular analysis whilst simultaneously yielding a measure of bronchial responsiveness. Ten normal subjects and a heterogenous group of 24 asthmatic patients (range % predicted FEV1 43.3-111.5) underwent HS challenge. Sputum samples induced were analysed. Total and differential cell counts between the two groups were compared. The association between bronchial responsiveness to HS and sputum cell counts was examined in the asthma group. Mean maximum fall in FEV1 for normal subjects was 4.0 (2.1-5.9, 95% CI)% after saline. Geometric mean PD20HS for asthma patients was 7.7 (range 0.68-40.92)ml. Adequate sputum samples were obtained from 9/10 normals and 23/24 asthmatic patients. Sputum from normal subjects contained a median of 3.8 (2.8-8.1, interquartile range)% eosinophils compared with 17.6 (8.9-34.1)% in sputum from asthma patients (P < 0.001). Sputum from asthma patients contained fewer of all other cell types compared with normals, with the difference in macrophages reaching significance. There was no correlation between PD20HS and cell count for any cell type in asthma subjects. Analysis of induced sputum represents a simple, safe, non-invasive and well-tolerated method of assessment of bronchial inflammation, suitable for use in patients with a range of asthma severity. There was no relationship between inflammation, as assessed by sputum cell counts and a measure of 'indirect' bronchial responsiveness.