Flavonoids are major components in a large number of medicinal herbs and showed inhibitory and stimulatory effects on the activation of aflatoxin (AF) B1 by liver microsomes. AFB1 oxidation was studied with purified recombinant human P450 1A2, 3A4, and 3A5 in reconstituted systems. Analysis of the AFB1 oxidation products showed that P450 1A2 formed AFM1, a small amount of AFQ1, and both exo- and endo-8,9-epoxides; P450 3A4 formed AFQ1 and exo-8,9-epoxide. P450 3A5, with 85% amino acid sequence identity to P450 3A4, formed mainly the exooxide and only a small amount of the detoxication product AFQ1. The exo epoxide is the most potent genotoxic and mutagenic metabolite of AFB1. The exo-oxide formation rates of 1A2 and 3A5 were - 14% and 43% of 3A4. These results coincide with the view that P450 3A4 is a major human liver P450 involved in AFB1 activation. A synthetic flavone, o-naphthoflavone ( oNF) caused potent inhibition of all oxidation reactions catalyzed by P450 1A2. In contrast to P450 1A2, addition of oNF to P450 3A4 resulted in a 2-fold increase of oxide formation but decreased AFQ1 formation. However, there was no obvious effect of AFB1 on the oxidation of oNF by P450 3A4 to its 5,6-oxide. Plots of AFQ1 and oxide formation rates vs AFB1 concentration were sigmoidal with Hill coefficients (n) of 2.4 and 4.1. Velocity vs AFB1 concentration plots were shifted to a more hyperbolic nature in the presence of oNF (n=l.4 for AFQ1, 1.6 for AFB1-oxide). This bidirectional modulatory effect was not changed by increasing P450 reductase/ P450 molar ratio. oNF showed the same modulatory effects in a P450 3A4/ NADPH-P450 reductase fusion protein system. The balance between 3o-hydroxylation (AFQ1) and epoxidation was influenced by changing the reduction system; flavodoxin-and ferredoxindependent systems yield only epoxidation. Also, only epoxidation was seen with P450 3A4 supported by the oxygen surrogates iodosylbenzene and cumene hydroperoxide. These results support a model that o NF affects AFB1 oxidation of P450 3A4 allosterically.