Emodin (1,6,8-trihydroxy-3-methylanthraquinone), an important aglycone found in natural anthraquinone glycosides frequently used in laxative drugs, was mutagenic in the Salmonella/mammalian microsome assay (Ames test) with a specificity for strain TA1537. The mutagenic activity was activation-dependent with an optimal amount of S9 from Aroclor 1254-treated male Sprague-Dawley rats of 20% in the S9 mix (v/v) for 10 μg emodin per plate. Heat inactivation of the S9 for 30 min at 60°C prevented mutagenicity. The addition of the cytochrome P-448 inhibitor 7,8-benzoflavone (18.5 nmoles per plate) reduced the mutagenic activity of 5.0 μg emodin per plate to about one third, whereas the P-450 inhibitor metyrapone (up to 1850 nmoles per plate) was without effect. To test whether a metabolite binds covalently to Salmonella DNA, [10- 14C]emodin was radiosynthesized, large batches of bacteria were incubated with [10- 14C]emodin and DNA was isolated. [G- 3H]Aflatoxin B 1 (AFB 1) was used as a positive control mutagen known to act via DNA binding. DNA obtained after aflatoxin treatment could be purified to constant specific activity. With emodin, the specific activity of DNA did not remain constant after repeated precipitations so that it is unlikely that the mutagenicity of emodin is due to covalent interaction of a metabolite with DNA. The antioxidants vitamin C and E or glutathione did not reduce the mutagenicity. Emodin was also negative with strain TA102. Thus, oxygen radicals are probably not involved. When emodin was incubated with S9 alone for up to 50 h before heat-inactivation of the enzymes and addition of bacteria, the mutagenic activity did not decrease. It is concluded that the mutagenicity of emodin is due to a chemically stable, oxidized metabolite forming physico-chemical associations with DNA, possibly of the intercalative type. In order to check whether an intact mammalian organism might be able to activate emodin to a DNA-binding metabolite, radiolabelled emodin was administered by oral gavage to male SD rats and liver DNA was isolated after 72 h. Very little radioactivity was associated with the DNA. Considering that DNA radioactivity could also be due to sources other than covalent interactions, an upper limit for the covalent binding index, CBI = (μmoles chemical bound per moles DNA nucleotides)/(mmoles chemical administered per kg body weight) of 0.5 is deduced. This is 10 4 times below the CBI of AFB 1. The demonstration of a lack of covalent interaction with DNA both in Salmonella and in rat liver is discussed in terms of a reduced hazard posed by emodin as a mutagenic drug in use in humans.