During the past ten years, the CASSCF/CASPT2 method has been applied with considerable success to a substantial number of problems related to the electronic spectroscopy of transition metal complexes, thus providing a definite breakthrough of ab initio quantum chemistry in this domain. This will be illustrated in the present contribution by means of a few representative cases from the field of inorganic, organometallic and bio-chemistry. Furthermore, CASPT2 results obtained for the excitation energies of the ions UO22+ and UO2Cl42- will be presented, indicating that the method is also applicable with comparable accuracy for molecules with very heavy metals ( provided that relativistic effects are accounted for). We will also show that the success of the method is critically dependent upon its judicious application, in particular upon the choice of the orbitals to be included in the reference active space. A link will be made between the latter choice and the specific electronic structure of the metal - ligand interactions.