32P-Postlabeling is a widely applied assay for the analysis of carcinogen-DNA adducts. Optimization of most steps in this assay has been given attention, but influences of DNA isolation and DNA purity on adduct quantitation have not been investigated systematically. In this study, DNA was isolated from human lymphocytes exposed to benzo[a]pyrene (B[a]P, 10 microM) for 18 hr and from liver of rats i.p.-treated with B[a]P (10 mg/kg body weight) using two different DNA isolation methods: a phenol-extraction and a salting-out procedure. Subsequently, DNA was analysed by nuclease P1 (NP1) or butanol-enriched 32P-postlabeling. Influences of RNA contamination were studied by labeling RNA isolated from in vitro exposed lymphocytes. In the in vitro experiment, DNA adduct levels were significantly higher using the salting-out procedure (63.2 +/- 13.7 adducts per 10(8) nucleotides, n = 9) as compared with the phenol-extraction (14.3 +/- 0.8). RNA was approximately 4 times less efficiently labeled as compared to DNA. Nonetheless, RNA contamination of DNA samples may result in an overestimation of DNA adduct levels when butanol enrichment is used, because RNA adduct levels seemed to be substantially higher than DNA adduct levels in the same cells. DNA adduct analysis by nuclease P1 enrichment is probably less affected, since RNA adducts appeared to be NP1 sensitive. In vivo, three different adducts were found by NP1 enriched 32P-postlabeling in the liver of B[a]P-exposed rats. Again, DNA adduct levels were significantly higher using salting out as compared to phenol extraction for the adduct which comigrated with the BPDE-DNA adduct standard (adduct 1) and an unknown adduct (adduct 2). However, the results were the opposite for another B[a]P-derived DNA adduct (adduct 3). Our results suggest that differences in DNA isolation procedures as well as RNA contamination influence quantitative DNA adduct analysis by 32P-postlabeling.