We investigated the applicability of induced sputum (IS), a non-invasive derivative from the lower respiratory tract, for smoking-related DNA adduct analysis and its comparability with peripheral blood lymphocytes (PBL), Lipophilic DNA adducts were quantified by the P-32-post-labeling assay in IS and PBL of smokers (n = 9) with stable smoking status at three time points tone week intervals) and nonsmokers (n = 9) at one time point. The success rate for sputum induction was 100% at all time points. There was no significant difference in total cell count, cell viability, squamous cell count and DNA yield between smokers and non-smokers. Within the smokers, there was no significant difference in IS cytology at the three time points: overall (mean of three measurements) total cell count, 9.0 +/- 2.4 x 10(6); cell viability, 77 +/- 4%; squamous cell count, 28 +/- 5%; non-squamous cell count, 72 +/- 4% (bronchoalveolar macrophages, 75 +/- 6%; neutrophils, 17 +/- 3%; bronchoepithelial cells, 7 +/- 2%; lymphocytes, 0.7 +/- 0.2%; metachromatic cells, 0.3 +/- 0.2%). IS DNA yield did not differ significantly at the three time points [overall (mean of three extractions) DNA yield, 66 +/- 20 mu g]. A typical smoking-associated diagonal radioactive zone was observed in the adduct maps of IS and PBL of all and five smokers, respectively, and of none of the non-smokers. Lipophilic DNA adduct levels in both IS and PBL of smokers were higher than those of non-smokers (3.7 +/- 0.9 versus 0.7 +/- 0.2/10(8) nt, P = 0.0005, and 2.1 +/- 0.3 versus 0.6 +/- 0.1/10(8) nt, P = 0.0001, respectively). In smokers the level of adducts in IS was non-significantly higher than that in PBL (3.7 +/- 0.9 versus 2.1 +/- 0.3/10(8) nt, P = 0.1), whilst in non-smokers the difference was not appreciable (0.7 +/- 0.2 versus 0.6 +/- 0.1/10(8) nt), Within the smokers there was no significant change in the level of adducts at the three time points either in IS or in PBL (coefficients of variation 34 and 29%, respectively). Adduct levels in IS at each time point were higher than those in PBL, leading to a significantly higher overall (mean of three quantifications) level of adducts in IS than PBL (3.3 +/- 0.2 versus 2.1 +/- 0.1/10s nt, P = 0.02). The overall levels of adducts in both IS and PBL were dose-dependently related to smoking indites. We conclude that IS is a preferable matrix as compared with PBL for molecular dosimetry of (current) exposure to inhalatory carcinogens as its analysis reveals both the existence and the magnitude of exposure more explicitly.