This paper describes an investigation into the general trend in electronic properties of anatase TiO2 photocatalysts co-doped with transition metals and nitrogen employing first-principles density functional theory. Fourteen different transition metals (M), including Sc, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Y, Zr, Nb, Mo, and Cd, have been considered. The characteristic band structures of the co-doping systems involving the transition metal series are presented. Our results indicate that the absorption edges of TiO2 are shifted to the visible-light region upon introduction of dopants, due to the reduced conduction band minimum (CBM) and the formation of impurity energy levels (IELs) in the band gap. These IELs are primarily formed from (a) the anti-bonding orbitals of the M-O (M indicates the doped transition metal) bonds, (b) the unsaturated nonbonding d orbitals of the doped transition metal (mainly d(xy), d(yz), and d(xz)), and (c) the Ti-O bonding/Ti-N anti-bonding orbitals of the bond next to the doped transition metal. When the valence d electrons of the doped metal are between 3 and 7, all three types of IELs appear in the band gap of the (M, N) co-doped systems. For systems doped with a metal of more than 7 valence electrons, only types (a) and (c) of IELs as well as the unoccupied pz state of N are observed. Based on our analysis, we propose that the co-doping systems such as (V, N), (Cr, N), and (Mn, N), which have the IELs with a significant bandwidth, are of great potential as candidates for photovoltaic applications in the visible light range.