The lung is a major target organ for smoking-associated cancer. We examined the applicability of induced sputum for molecular dosimetry of exposure to tobacco smoke-related carcinogens. Sputum induction was performed by inhalation of 4.5% saline delivered from an ultrasonic nebulizer for a period of up to 21 min in a group of smoking (n = 20) and nonsmoking (n = 24) healthy individuals. Samples were analyzed for total and differential cell counts and cell viability. Subsequently, DNA contents of the samples were isolated, and measurement of lipophilic DNA adducts was done by the 32P-postlabeling assay using nuclease P1 (NP1) and butanol enrichment methods. All subjects tolerated the induction procedure without experiencing any troublesome symptoms, and 90% of smokers (18 of 20) and 88% of nonsmokers (21 of 24) succeeded in producing sufficient amounts of sputum. Total cell counts and percentages of viable cells in smokers were higher than those in nonsmokers (6.7+/-6.0 versus 4.7+/-6.0 x 10(6), P = 0.40 and 80+/-15 versus 63+/-17, P = 0.01, respectively). In cell differentials, smokers had lower percentages of bronchoalveolar macrophages and higher percentages of neutrophils (69+/-24 versus 92+/-5, P = 0.002 and 26+/-26 versus 4+/-4, P = 0.008, respectively). Using the NP1 digestion method, all smokers and only one nonsmoker showed a diagonal radioactive zone in their adduct maps; adduct levels in smokers were higher than those in nonsmokers (3.1+/-1.4 versus 0.6+/-0.8/10(8) nucleotides; P = 0.0007), and also, adduct levels were significantly related to smoking indices. Applying the butanol extraction method, however, only half of the smokers and three nonsmokers showed the diagonal radioactive zone in their adduct maps; adduct levels in smokers were higher than those in nonsmokers (4.6+/-3.7 versus 1.0+/-1.9/10(8) nucleotides; P = 0.02), and the levels of adducts were significantly related to the smoking indices. There was a correlation between the levels of adducts determined by the two enrichment methods (r = 0.7; P = 0.02). Paired comparison showed no differences between the levels of adducts measured by the two methods (P = 0.55). We conclude that induced sputum can serve for molecular dosimetry of inhalatory exposure to carcinogens and that the NP1 version of the 32P-postlabeling assay is a choice of preference for studying smoking-induced DNA adducts in the lower respiratory tract.