Inhalation of hypertonic saline to induce sputum may alter cells and fluid-phase markers in sputum. We have compared indices of inflammation in sputum produced spontaneously with sputum induced by an aerosol of hypertonic saline. Twenty-three asthmatics produced spontaneous followed by induced sputum on the same day. The sputum specimen was separated from saliva within 2 h, dispersed with dithiothreitol (DTT) and processed to obtain cytospins and supernatant. The statistical power to detect a 20% difference in sputum parameters was > 90%. Results are expressed as median and interquartile range [IQR]. Induced sputum had a higher proportion of viable cells (77.0 [19.0] versus 47.0 [38.0]%, p < 0.001), less squamous cell contamination (1.0 [1.2] versus 1.8 [34.0]%, p < 0.001) and better quality cytospins (score of 8.0 [4.0] versus 4.0 [2.0], p < 0.001). It also had lower fluid-phase levels of eosinophil cationic protein (ECP) (1,358 [1,102] versus 1,574 [2,479] microg/L) and fibrinogen (1,560.0 [3,130.0] versus 4,350.0 [5,970.0] ng/ml) but only the latter was significantly different (p = 0.02). Induced sputum was similar to spontaneous sputum in weight (200.0 [219.0] versus 270.0 [227.0] mg), total cell count (3.3 [4.1] versus 3.5 [4.5] x 10(6)/ml), proportion of nonsquamous cells, and levels of tryptase. The agreement between induced and spontaneous measurements was good, but fluid-phase levels were affected by the low viability of some spontaneous samples. We conclude that for the indices measured in asthmatic subjects, induced sputum separated from saliva is similar to lower respiratory secretions expectorated spontaneously and has the advantage of better cell viability.