Selective catalytic reduction (SCR) of NO via NH3 and O-2 over the V2O5 surface has been the focus of considerable research interest due to its role in mitigating air pollution. Our theoretical investigations at the periodic DFT level reveal that the Lewis acid active center could be a starting point in the dominant Eley-Rideal mechanism, while other active sites might either exist or be formed during the reaction process and play roles in competition. In this systematic study, an integrated catalytic cycle consisting of four module steps (i) NH3 + NO + V2O5 -> N-2 + H2O + HV2O5, (ii) NH3 + NO + HV2O5 -> N-2 + H2O + HHV2O5, (iii) NH3 + NO + HHV2O5 -> N-2 + 2H(2)O + HV2O4, and (iv) NH3 + NO + O-2 + HV2O4 -> N-2 + 2H(2)O + V2O5 is proposed by using uniform theoretical model for the most possible processes involved. This suggested mechanism is easy to understand and agrees well with the experimental observations and results of other theoretical studies. More satisfactory, differences in the catalytic activity for diverse active sites can be explained not only by relative energies and barrier heights but also by geometries of the intermediates and transition states appeared in the cycle. For Step I, the formation of species HO-V-NH2 followed by H-migration of HO at Lewis acid site V-a is decisive because of the very high activation energy (63.6 kcal/mol), while following transformations and the release of N-2 and H2O are relatively easy. The most favorable path is however going through V-b site for Step I. The change from intermediate 1 to 2 must suffer a barrier of 52.7 kcal/mol, which is only 10.9 kcal/mol lower than that for V-a. After the formation of one V-OH Bronsted acid site, the transformation from intermediate 11 to 12 is the most difficult process for Step II (E-a = 38.6 kcal/mol). The most stable configuration for double V-OH sites displays two potential pathways depending on the priority of removing H2O at the late stage of Step III. Our calculations indicate that Step IV favors to occur through HVt1 related pathway in which the oxygen vacancy and V-OH sites are opposite to each other. This novel multi-step mechanism can provide us a deeper understanding of the SCR reaction over V2O5 surface, and we expect that the design of SCR catalyst could be improved on the basis of theoretical predictions related to these key sites and important processes. (C) 2013 Elsevier Inc. All rights reserved.