DNA polymerases are crucial constituents of the complex cellular machinery for replicating and repairing DNA. Discerning mechanistic pathways of DNA polymerase on the atomic level is important for revealing the origin of fidelity discrimination. Mammalian DNA polymerase beta (pol beta), a small (39 kDa) member of the X-family, represents an excellent model system to investigate polymerase mechanisms. Here, we explore several feasible low-energy pathways of the nucleotide transfer reaction of pol beta for correct (according to Watson-Crick hydrogen bonding) G:C basepairing versus the incorrect G:G case within a consistent theoretical framework. We use mixed quantum mechanics/molecular mechanics (QM/MM) techniques in a constrained energy minimization protocol to effectively model not only the reactive core but also the influence of the rest of the enzymatic environment and explicit solvent on the reaction. The postulated pathways involve initial proton abstraction from the terminal DNA primer O3'H group, nucleophilic attack that extends the DNA primer chain, and elimination of pyrophosphate. In particular, we analyze several possible routes for the initial deprotonation step: (i) direct transfer to a phosphate oxygen O(Palpha) of the incoming nucleotide, (ii) direct transfer to an active site Asp group, and (iii) transfer to explicit water molecules. We find that the most probable initial step corresponds to step (iii), involving initial deprotonation to water, which is followed by proton migration to active site Asp residues, and finally to the leaving pyrophosphate group, with an activation energy of about 15 kcal/mol. We argue that initial deprotonation steps (i) and (ii) are less likely as they are at least 7 and 11 kcal/mol, respectively, higher in energy. Overall, the rate-determining step for both the correct and the incorrect nucleotide cases is the initial deprotonation in concert with nucleophilic attack at the phosphate center; however, the activation energy we obtain for the mismatched G:G case is 5 kcal/mol higher than that of the matched G:C complex, due to active site structural distortions. Taken together, our results support other reported mechanisms and help define a framework for interpreting nucleotide specificity differences across polymerase families, in terms of the concept of active site preorganization or the so-called "pre-chemistry avenue".