The formation and repair of benzo[a]pyrene diol epoxide-N2-deoxyguanosine adducts (BPDE-N2-dG) in DNA isolated from the skin of mice treated topically with benzo[a]pyrene (BP) was studied by 32P-postlabeling and by low-temperature fluorescence spectroscopy under low resolution and under high resolution fluorescence line narrowing (FLN) conditions. In agreement with earlier studies, total BP-DNA binding reached a maximum at 24 h after treatment (dose: 1 mumol/mouse), then declined rapidly until 4 days after treatment and much more slowly thereafter. An HPLC method was developed which resolved the 32P-postlabeled (-)-trans- from (-)-cis-anti-BPDE-N2-dG, and (+)-trans-from (+)-cis-anti-BPDE-N2-dG. High performance liquid chromatography analysis of the major TLC adduct spot (containing > 80% of the total adducts) obtained by postlabeling BP-modified mouse skin DNA showed that it consisted of a major component that coeluted with (-)-cis-/(+)-trans-anti-BPDE-N2-dG and a minor component that coeluted with (-)-trans-/(+)-cis-anti-BPDE-N2-dG and that the minor component was repaired at a slower rate than the major component. Low-temperature fluorescence spectroscopy of the intact DNA identified the major adduct as (+)-trans-anti-BPDE-N2-dG and the minor adduct fraction consisted mainly of (+)-cis-anti-BPDE-N2-dG. In agreement with the 32P-postlabeling results it was observed by fluorescence spectroscopy that the (+)-cis-adducts were repaired more slowly than most other adducts. Moreover, the (+)-trans-adducts exhibited a broad distribution of base-stacked, partially base-stacked and helix-external conformations. Mouse skin DNA samples obtained at early timepoints (2-8 h) after treatment with BP contained substantially more of the 'external' adducts, while samples at later timepoints (24-48 h) contained relatively more adducts in the base-stacked conformation, indicating also that the latter adducts are repaired less readily than the former. The possible biological significance of these novel observations of conformation-dependent rates of DNA adduct repair and their possible dependence on DNA sequence, are discussed.