The geometries of the hexacarbonyls and pentacarbonyls of chromium, molybdenum, and tungsten are optimized at the Hartree-Fock and MP2 levels of theory using effective core potentials for the metal atoms. The M-CO bond lengths of Mo(CO)(6) and W(CO)(6) predicted at the MP2 level using moderate valence basis sets are in excellent agreement with experimental values. The Cr-CO bond length in Cr(CO)(6) calculated at MP2 is too short. The total bond energies of the metal hexacarbonyls calculated at the CCSD(T) level of theory are slightly lower than the experimentally derived values. The first dissociation energies calculated at CCSD(T) using MP2-optimized geometries for M(CO)(6) and M(CO)(5) are in very good agreement with experimental results for Mo(CO)(6) and W(CO)(6) from gas-phase laser pyrolysis. The calculated first dissociation energy at CCSD(T) for Cr(CO)(6) using the MP2-optimized geometries for Cr(CO)(6) and Cr(CO)(5) is too high. The theoretical and experimental results suggest the following first dissociation energies Delta H-298 for the M(CO)(6) compounds : Cr(CO)(6) = 37 +/- 2 kcal/mol; Mo(CO)(6) = 40 +/- 2 kcal/mol; W = 46 +/- 2 kcal/mol. The agreement of previously reported theoretic;al dissociation energies using density functional theory with kinetic data for the activation energy of substitution reactions showing a different order for the hexacarbonyls Mo < Cr < W is misleading. The kinetic data for Mo(CO)(6) and W(CO)(6) refer to a different mechanism and should not be used to estimate the metal-carbonyl bond strength.