In a continuing systematic project investigating complexes of second-row transition metals, results are presented here for different types of systems where metal-oxygen bonds are present. The entire sequence of second-row dioxides and some trioxides and the single tetraoxide of interest, RuO4, have been studied. An interesting result for the dioxides is that most of these systems are bent, which is in contrast to the corresponding dichloride systems. The origin of the bent structures of the dioxides is a more optimal mixing between metal 4d-orbitals and oxygen orbitals for bent than for linear geometries. This type of mixing is a general effect that also influences the structure of other transition metal complexes, such as MoH6, which obtains a C3nu structure. Complexes involving molecular oxygen as a ligand have also been studied. Two types of structures occur, the symmetric peroxide low-spin structures and the superoxide high-spin structures which are bound to the metal by a single covalent bond. The superoxide structures can be eta1- or eta2-coordinated to the metal. Comparisons are made to previous calculations on transition metal oxides, and it is shown that the present higher level of accuracy is of large importance even for the qualitative chemistry of these systems.