Aristolochic acid I (AAI) and aristolochic acid II (AAII), the two major components of the carcinogenic plant extract aristolochic acid (AA), are known to be mutagenic and to form DNA adducts in vivo. According to the structures of the major DNA adducts identified in animals and humans, nitroreduction is the crucial pathway in the metabolic activation of these naturally occurring nitroarenes to their ultimate carcinogenic species. Using the nuclease P1-enhanced version of the 32P-post-labelling assay we investigated the formation of DNA adducts by AAI and AAII in different in vitro activation systems in order to determine the most suitable in vitro system mimicking target tissue activation. Although DNA adducts resulting from oxidative activation of AAs have not yet been identified both reductive and oxidative in vitro systems were employed. In vitro incubations were conducted under standardized conditions (0.3 mM AAs; 4 mM dNp as calf thymus DNA) using rat liver microsomes, xanthine oxidase (a mammalian nitroreductase), horseradish peroxidase, lactoperoxidase and chemical reduction by zinc. Enzymatic incubations were performed under aerobic and anaerobic conditions. A combination of two independent chromatographic systems (ion-exchange chromatography and reversed-phase HPLC) with reference compounds was used for the identification of DNA adducts detected by the 32P-post-labelling assay. The two known major adducts of AAI or AAII found in vivo were generated by all in vitro systems except for incubations with AAII and horseradish peroxidase where two unknown adducts predominated. Irrespective of the in vitro activation system used, the majority of adduct spots obtained were identified as the previously characterized four AA-DNA adducts: dA-AAI, dA-AAII, dG-AAI and dG-AAII. This indicates that both reductive and peroxidative activation of AAI or AAII resulted in chromatographically indistinguishable DNA adducts. Thus, peroxidase mediated activation of AAs led to the formation of the same adducts that had been observed in vivo and upon reductive activation in several in vitro systems. Quantitative analyses of individual adducts formed in the various in vitro systems revealed relative adduct labelling (RAL) values over a 100,000-fold range from 4 in 10(3) for activation of AAII to deoxyadenosine adducts by zinc to only 3 in 10(8) for activation of AAII by lactoperoxidase. The extent of DNA modification by AAI was higher than by AAII in all enzymatic in vitro systems. Only activation by zinc resulted in higher total binding to exogenous DNA by AAII than by AAI. Aerobic incubations with rat liver microsomes generated AAI- and AAII-DNA adduct profiles reproducing profiles in target tissue (forestomach) of rats, thus providing the most appropriate activation among the in vitro systems tested.