Recent CASPT2 calculations of the heats of formation of the isomeric benzynes by R. Lindh and M. Schütz [Chem. Phys. Lett. 258 (1996) 409] are re-examined. The unrealistically low value reported for p-benzyne (132.7 kcal/mol) is shown to be an artifact of the use of incorrect CASSCF and CASPT2 energies for p-benzyne, as well as a flawed isodesmic reaction analysis. Use of the correct energies and an appropriate isodesmic reaction leads to excellent agreement between the calculated and measured heats of formation for p-benzyne. The performance of coupled-cluster methods and density functional theory in predicting benzyne thermochemistry and singlet-triplet splittings is also evaluated.