To determine whether induced sputum samples might provide a useful means for evaluating the effects of therapy on airway mucosal inflammation, we examined induced sputum samples obtained before and after 6 days of treatment with prednisone (0.5 mg/kg/day) or placebo in a randomized, double-blind study of 24 asthmatic subjects. Induced sputum was analyzed for total and differential cell counts and for concentrations of eosinophil cationic protein, albumin, and mucin-like glycoprotein. We found that the mean (+/- SEM) percentage of eosinophils in sputum samples from the prednisone-treated group fell from 14.1% +/- 5.0% at baseline to 1.8% +/- 0.8% after treatment, a decrease significantly greater than in the placebo-treated group (from 10.3% +/- 4.9% to 11.1% +/- 4.0%; p = 0.002). The absolute number of eosinophils also decreased significantly more in the prednisone-treated group than in the placebo-treated group (p = 0.04). In addition, eosinophil cationic protein levels in induced sputum fell more in the prednisone-treated group than in the placebo-treated group (from 324 +/- 131 ng/ml to 144 +/- 84 ng/ml vs 173 +/- 50 ng/ml to 188 +/- 47 ng/ml; p = 0.002). Furthermore, prednisone treatment was associated with a significant increase in peak expiratory flow, an effect that was significantly correlated with the decrease in eosinophil percentage in induced sputum (rs = 0.64, p = 0.04). Prednisone treatment was not associated with any significant change in the concentrations of albumin or mucin-like glycoprotein. We conclude that analysis of induced sputum is a useful noninvasive method for studying the effects of asthma therapy on airway eosinophilic inflammation.