The possibilities for using CaNb2O6 as a photocatalyst in direct water splitting have been evaluated by investigating the electronic structure of the material. In addition the oxide was doped with nitrogen in order to modify the electronic structure and obtain visible light absorption. Experimental techniques such as electrochemical impedance spectroscopy (EIS), photocurrent, and diffuse reflectance spectroscopy (DRS) were combined with theoretical approaches to determine the bandgap, flatband potential and quasi-Fermi levels of the photocatalyst. CaNb2O6 was prepared by a sol-gel synthesis and doped with nitrogen by heat treatment of the oxide powder in an ammonia atmosphere. X-ray diffraction (XRD) confirmed phase pure orthorhombic CaNb2O6 for both pure and N-doped oxide and excluded a possible transformation of the oxide into an oxynitride. Upon illumination anodic photocurrents were observed implying that CaNb2O6 was an n-type semiconductor due to oxygen vacancies in the lattice. From the wavelength dependency of the photocurrent a direct bandgap of 3.7eV and an indirect bandgap of 3.4eV were determined for undoped CaNb2O6. Doping with nitrogen altered the optical properties of the oxide and shifted the absorption edge into the visible light region. Calculations using the density functional theory (DFT) attributed the change in absorption properties to the formation of narrow energy bands above the valence band of pure CaNb2O6. An alternative explanation could be a hybridization of N 2p and O 2p bands. Correspondingly a reduction of the bandgaps for N-doped CaNb2O6 with respect to the undoped oxide was identified. Impedance was applied to determine the flatband potential of CaNb2O6 from Mott-Schottky plots. However the obtained results seemed to be dominated by contributions from the electrode substrate. Theoretical investigations concluded that pinhole-free oxide layers creating an ohmic contact with the substrate are required in order to designate the observed impedance response to the space charge capacitance. Quasi-Fermi level measurements indicated a low photocatalytic activity of CaNb2O6 as no photocurrent could be detected. Further investigations are needed to identify the cause of the photocurrent limitations. Nevertheless probable explanations could be low conductivity in CaNb2O6, high concentrations of recombination centers or slow charge transfer kinetics. The latter was confirmed for porous oxide layers as the addition of a hole scavenger increased the measured photocurrent. Positive photocurrent transients were also observed for porous CaNb2O6 films and could be related to either the diffusion of electrons through the porous oxide layer or to a photoanodic decomposition of the photocatalyst.