The method that we have previously reported for sputum induction involves collecting the entire expectorate produced over a 20 min inhalation of 3% saline aerosol. This method presents the potential disadvantage of a considerable and variable salivary contribution to the induced sputum sample. In this study, we examined whether separate collection of saliva and sputum represents a better method for collecting induced sputum during sputum induction. In 11 stable asthmatics, we compared the volume, total and differential cell counts, and eosinophil cationic protein (ECP) levels in four induced sputum samples, two performed using our previous method (Method A) and two using another method (Method B) in which subjects spit saliva into one container before coughing sputum into another. We found that the volume of sputum obtained with Method B was lower than that obtained with Method A (6.16 +/- 0.61 vs 20.1 +/- 2.7 mL; p = 0.003), as was the percentage of squamous cells (34 +/- 4 vs 47 +/- 6; p = 0.023). In addition, the ECP levels in samples collected by Method B were higher (261 +/- 42 vs 145 +/- 26 ng.mL-1; p = 0.01). The differential counts of nonsquamous cells were similar except for the percentage of neutrophils, which was lower in Method B (37 +/- 4 vs 50 +/- 5%; p = 0.019). The repeatability of measurements of eosinophil percentages and of ECP levels was similar for the two methods. We conclude that separate collection of saliva and sputum yields induced sputum samples with reduced amounts of saliva and is, therefore, a better method for collecting induced sputum.