The calculation of accurate bond strengths for transition metal complexes is discussed. Emphasis is put on methods capable of treating relatively large complexes. Results from parametrized ab initio schemes (PCI-80) and hybrid density functional meth ods (B3LYP) are compared to accurate experiments for se quences of small transition metal complexes. In most cases both methods give quite satisfactory results with the PCI-80 results slightly better in general. However, for a few cases where the correlation effects are unusually large, the PCI-80 scheme breaks down. For large systems the B3LYP method is the most promis ing approach since it is very fast and sufficiently accurate for most problems.The calculation of accurate thermodynamical properties of transition metal complexes is one of the most difficult problems in computational quantum chem istry. First, the calculation of accurate bond strengths for any molecule, even for a small first row molecule, is difficult, requiring methods that account for almost all of the valence correlation energy. As a simple rule of thumb, the rel ative error in the correlation effect on a bond strength will be the same as the relative error in the total valence correlation energy. Secondly, transition metal complexes are generally accepted to be more difficult to treat than molecules composed of lighter atoms. From an ab initio viewpoint, there can be a very strong coupling between near degeneracy effects and large dynamical correlation effects for transition metal complexes. From a density functional theory (DFT) viewpoint, one difficulty is the strong coupling between correlation effects and exchange effects, at least for cases with unfilled d-shells. To these two difficulties can be added the problem that even the smallest transition metal complexes