Sputum examination is a useful noninvasive method to study airway inflammation. We investigated the reproducibility and validity of the measurements of lymphocyte subsets in the sputum of 11 stable patients with asthma and 10 nonasthmatic smokers. Sputum was dispersed with 0.1% dithiothreitol. A differential cell count was made with Wright's stain. Aliquots were stained with antibodies to CD19 (B cells), CD3 (T cells), CD4 (helper), CD8 (suppressor), and the activation marker CD25 (IL2 receptor) on T-cell subsets and were assayed by flow cytometry. Sputum from patients with asthma compared with nonasthmatic subjects had more eosinophils (mean +/- SEM, 32.5 +/- 8.5 versus 1.3 +/- 0.5%, p < 0.01) and a higher proportion of lymphocytes that were B cells (16.2 +/- 3.2 versus 4.0 +/- 1.0%, p < 0.01), and these correlated closely with the eosinophils (r = 0.8, p < 0.01). Patients with asthma also had more activated T-helper cells (39.3 +/- 4.6 versus 9.0 +/- 9.0%, p = 0.05), but the comparison was limited to two smokers because of macrophage autofluorescence. The repeatability of measurements of helper T-cells (R = 0.94), suppressor T cells (R = 0.88), and activated helper T cells (R = 0.77) was good; repeatability of measurements of T and B cells could not be examined because these were reciprocals of each other. Asthmatic sputum has different lymphocyte profiles than sputum from nonasthmatic smoking control subjects. The results demonstrate a potential importance of antibody-producing lymphocytes in the airway and their relation to sputum eosinophilia in asthma.