The biosynthesis of the potent environmental carcinogen aflatoxin B1 involves ca. 15 steps beyond the first polyketide intermediate. Central among these is the rearrangement of the anthraqinone versicolorin A to the xanthone demethylsterigmatocystin. Genetic evidence strongly suggests that two enzymes are required for this process, a cytochrome P450, AfIN, and a probable NADPH-dependent oxidoreductase, AfIM. Given the overall redox change evident in this skeletal rearrangement, two rounds of oxidation and a reduction necessarily occur. Earlier experiments indicated that reductive deoxygenation of versicolorin A is not the first step. In the present report we consider a mechanistic alternative that AfIM-mediated reduction is instead the last of these three reactions prior to formation of the xanthone intermediate. To this end, 9-hydroxydihydrodemethylsterigmatocystin was prepared by total synthesis as was its 9-deoxy analogue, an established aflatoxin precursor. During the final isolation of the "angular" synthetic xanthone targets it was found that acid catalysis promoted their isomerization to thermodynamically favored "linear" xanthones. Whole-cell and ground-cell incubations of the 9-hydroxy- and 9-deoxyxanthones were conducted with a mutant strain of Aspergillus parasiticus blocked at the first step of the pathway and examined for their ability to support aflatoxin production. The 9-deoxyxanthone gave dramatically enhanced levels of the mycotoxin. The 9-hydroxyxanthone, on the other hand, afforded no detectable increase in aflatoxins above controls, indicating that reductive deoxygenation at C-9 of a xanthone precursor does not take place in aflatoxin biosynthesis. Constraints imposed by earlier studies and the experiments in this paper serve to eliminate simple and intuitive conversions of versicolorin A to demethylsterigmatocystin and lead inescapably to a more subtle reaction sequence of oxidation-reduction-oxidation. Previous puzzling observations of extensive A-ring hydrogen exchange in the course of the rearrangement of versicolorin A to demethylsterigmatocystin have now been explained by a new mechanism that is consistent with all extant data. We propose that P450-mediated aryl epoxidation (AfIN) initially disrupts the aromatic A-ring of versicolorin A. Oxirane opening enables A-ring proton exchange, as does the subsequent AfIM-mediated reductive step. A second cycle of P450 oxidation (AfIN), this time a Baeyer-Villiger cleavage, enables decarboxylation and the formation of demethylsterigmatocystin. Mechanistic and stereoelectronic principles that underlie this proposal are described and may prove general as illustrated in biogenetic hypotheses for four other fungal anthraquinone &DRARR; xanthone transformations.