Using first principles quantum mechanics [nonlocal density functional theory (B3LYP)], we calculated the 13 most likely intermediate species for methanol oxidation on clusters of all 2nd and 3rd row Group VIII transition metals for all three likely binding sites (top, bridge, and cap). This comprehensive set of binding energies and structures allows a detailed analysis of possible reaction mechanisms and how they change for different metals. This illustrates the role in which modem quantum chemical methods can be used to provide data for combinatorial strategies for discovering and designing new catalysts. We find that methanol dehydrogenation is most facile on Pt, with the hydrogens preferentially stripped off the carbon end. However, water dehydrogenation is most facile on Ru. These results support the bifunctional mechanism for methanol oxidation on Pt-Ru alloys in direct methanol fuel cells (DMFCs). We find that pure Os is capable of performing both functionalities without cocatalyst. We suggest that pure Os be examined as a potential catalyst for low overpotential, highly dispersed catalyst DMFCs. Pathways to form the second C-O bond differ between the pure metals (Pt and Os) in which (CO)(ads) is probably activated by (OH)(ads) and the Pt-Ru binary system in which (COH)(ads) is probably activated by O-ads. For all cases we find that formation of (COOH)(ads) is an important precursor to the final dehydrogenation to desorb COP from the surface.