Completely optimized ab initio calculations at the Hartree-Fock and second-order Moller-Plesset (MP2) levels using basis sets up to D95++** and semiempirical calculations using AM1, PM3, and SAM1 are reported for acetylacetone and both the cis and trans isomers of its 1- and 2-enols. Comparison of the energies (MP2/D95** used for examples) of the various species indicates that the internal H-bond of the cyclic, conjugated cis-2-enol stabilizes that species by 12.0 kcal/mol, which is considerably more than for the cyclic 1-enol (6.0 kcal/mol). Furthermore, as the enhanced stability of the H-bond is more than for other resonance-stabilized H-bonds, the existence of aromatic character in this 6-pi-electron system is supported. The non-H-bonded trans-2-enol is more stable than the trans-1-enol by 5.0 kcal/mol, which we attribute to conjugation. The symmetric C-2 upsilon structure of the cis-2-enol is predicted to be 2.5 kcal/mol less stable than the unsymmetrical structure by both MP2/D95** and MP2/D95++(d,p) before the zero-point vibrational energy correction and 0.2 kcal/mol less than the unsymmetrical structure after correction (using scaled HF/D95** frequencies and the MP2/D95++** energies). The calculated structures of the cis-2-enol agree reasonably well with that reported by Karle but are not consistent with that reported by Shibata. Both Hartree-Fock (because of the electron correlation error) and the three semiemperical methods are inadequate for this problem, as they all give artificially high energies for the C-2 upsilon structure. We suggest that this structure may still be overestimated at the highest levels calculated due to incomplete correction for electron correlation which favors this structure.