To investigate subject-to-subject variability in commonly used parameters from bronchoalveolar lavage (BAL), we analyzed the results of BAL from 78 consecutive normal volunteer subjects. The BAL was performed using three, 40 ml aliquots of normal saline solution, in a lingular subsegment. The same diameter bronchoscope (4.8 mm) was used and BAL was performed and analyzed identically in all subjects. While the percentage of lavageate returned was relatively consistent from subject to subject (63.4 +/- 10.8 percent, mean +/- SD), there was marked variability in the number of cells obtained (7.3 +/- 3.9 X 10(6)). While low returns were generally associated with a diminished cell recovery, in the range of normal percentage returned, there was no correlation with the number recovered. There was no correlation of cell number or percent lavageate returned with gender, height, or any lung volume (as determined on pulmonary function tests). There was relatively little variability in differential analysis of recovered cells. The BAL was well tolerated by all subjects; the chest roentgenogram was found to be of little utility in screening normal subjects, but pulmonary function testing detected obstructive lung disease in three potential subjects leading to their disqualification as "normal." Our experience confirms the safety of bronchoalveolar lavage in normal volunteer subjects. We report the BAL results that can be expected for research or clinical comparison using a protocol described in detail. The variability of cell numbers from BAL, even under these controlled conditions, precludes the routine use of this measurement from BAL performed for clinical purposes.