The 3d-element transition metal dioxide MO(2), peroxide M(O(2)), and superoxide MOO clusters (M=Sc-Zn), are studied by density functional theory with the B1LYP functional. The reliability of the methods and basis sets employed was tested by a reinvestigation of the monoxides, for which a database of experimental data is available. The global minima on the M+O(2) potential energy surfaces correspond to dioxide structure, the only exception being CuOO, with a superoxide structure. All Zn dioxygen clusters are thermodynamically unstable-their ground states lie higher than the dissociation limit to Zn+O(2). Our calculations are in favor of the high-spin configurations for the FeO(2), CoO(2), and NiO(2) ground states, which are still a subject of extensive theoretical and experimental studies. These assignments are confirmed by the coupled-cluster method, CCSD(T), except for NiO(2). Based on the existence of a stable NiO(2) monoanion in a (4)B(1) state, however, it can be concluded that NiO(2) in its (5)A(1) state should also be stable. The vibrational frequencies are calculated for clusters entrapped in the cubic cell of solid Ar matrix and compared with those obtained for gas-phase clusters. The matrix has no influence on the vibrations of the monoxides and most of the dioxides; however, Co and Ni-dioxoclusters interact strongly with the atoms from the noble gas matrix. The most intense frequencies in the IR spectra are shifted to lower energies and the ordering of the low-lying electronic states by stability is also reversed. According to the electrostatic potential maps, the oxygen atoms in the peroxides are more nucleophilic than those in the dioxides and superoxides. The terminal oxygen atom in superoxides is more nucleophilic than its M-bonded oxygen atom, though charge distribution analysis predicts a smaller negative charge on the terminal oxygen. TiO(2) is the only dioxide in which nucleophilic character in the vicinity of the metal cation is induced.