Physically reasonable electronic structures of reconstructed rutile TiO2(110)-(1×2) surfaces were studied using density functional theory (DFT) supplemented with Hubbard U on-site Coulomb repulsion acting on the d electrons, so called as the DFT+U approach. Two leading reconstruction models proposed by Onishi-Iwasawa and Park et al. were compared in terms of their thermodynamic stabilities.PACS numbers: 71.15. Mb, 68.47.Gh Rutile TiO 2 and its surfaces represent model systems to explore the properties of transition metal oxides that are important in technological applications such as catalysis, photovoltaics, and gas sensing [1], to name a few. Truncated or stoichiometric (110) surface of rutile is the most stable one among all surfaces of titania [2]. Upon thermal annealing or ion bombardment TiO 2 (110)-(1×1) surface is reduced by loosing the bridging oxygens, and is often undergo a (1×2) reconstruction with row formations [3][4][5][6][7][8][9][10][11][12][13]. The identification of these rows on reconstructed surfaces, in three dimensions, is difficult by experimental methods [13]. The best candidate for modeling this reconstruction involves the addition of "Ti 2 O 3 " molecule on the surface unit cell (added-row model) proposed by Onishi and Iwasawa [3]. In addition to theoretical studies it was supported by electron stimulated desorption of ion angular distribution (ESDIAD), scanning tunnelling microscopy (STM), and low-energy electron diffraction (LEED) experiments [5][6][7][8][9][10][11]. On the other hand, Shibata et al. Now there is a debate about which formation gives rise to the (1×2) long range order, the last proposed one or the best previous candidate? In this context we study the electronic properties of these two models using Hubbard U corrected total energy density functional theory (DFT+U ) calculations to get physically reasonable results comparable to existing experimental data. We discuss which of these leading models can be assigned to describe the (1×2) reconstructed surface by comparing them according to their thermodynamic stabilities.Band structures of reduced and reconstructed TiO 2 (110) surfaces determined by pure DFT calculations does not agree with experiments [10,14,15]. Failure of the standart DFT is not limited to band-gap underestimation stemming from the many-electron self-interaction error (SIE). More importantly, it does not predict experimentally observed gap states [16,17] that are associated with the excess electrons due to the formation of surface oxygen vacancies. In DFT calculations, these Ti 3d electrons occupy the bottom of the conduction band (CB) giving metallic character. Hence, hybrid DFT methods need to be used. For instance, SIE can be partly corrected by partially mixing nonlocal Fock exchange term with DFT exchange term [18,19] or DFT+U approach [20] can make up for the lack of strong correlation between the 3d electrons, a shortcoming of common exchange-correlation functionals. Empirical Hubbard U term accounts for the on-site Coulomb repulsion between th...