We present a comparison of the following prominent propylene epoxidation mechanisms using H2O2/TS-1 at a consistent density functional theory (DFT) method: (1) the Sinclair and Catlow mechanism on tripodal site through Ti-OOH species, (2) the Vayssilov and van Santen mechanism on tetrapodal site without Ti-OOH formation, (3) the Munakata et al. mechanism involving peroxy (Ti-O-O-Si) species, (4) the defect site mechanism with a partial silanol nest, and ( 5) the defect site mechanism with a full silanol nest. We have reproduced the previously published (ethylene epoxidation) pathways (1-3) for propylene epoxidation using larger and SiH3-terminated cluster models of the T-6 crystallographic site of TS-1. Mechanism 5 is a new mechanism reported here for the first time. The use of a consistent level of theory for all the pathways allows for the first time a meaningful comparison of the energetics representing the aforementioned pathways. We have rigorously identified the important reaction intermediates and transition states and carried out a detailed thermochemical analysis at 298.15 K and 1 atm. On the basis of the Gibbs free energy of activation, the Sinclair and Catlow mechanism (Delta G(act) = 7.9 kcal/mol) is the energetically most favorable mechanism, which is, however, likely to operate on the external surface of TS-1 due to the tripodal nature of the Ti site in their model. The newly reported defect site mechanism (with a full silanol nest) is a competitive propylene epoxidation mechanism. There are two main steps: (1) hydroperoxy formation (Delta G(act) = 8.9 kcal/mol) and (2) propylene epoxidation (Delta G(act) = 4.6 kcal/ mol). This mechanism is likely to represent the chemistry occurring inside the TS-1 pores in the liquid-phase epoxidation (H2O2/TS-1) process and could operate in direct gas-phase epoxidation (H-2/O-2/Au/TS-1) as well. If only the propylene epoxidation step is considered, then the Munakata peroxo intermediate (Si-O-O-Ti) is the most reactive intermediate, which can epoxidize propylene with a negligible activation barrier. However, formation of the Munakata intermediate is a very activated step (Delta G(act) = 19.8 kcal/ mol). We also explain the trends in the activation barriers in different mechanisms using geometric and electronic features such as orientation of adsorbed H2O2 and propylene, hydrogen bonding, O1-Ti bond distance in the Ti-O1-O2-H intermediate, and O1-O2 stretching in the transition state. Implications of different Ti site models are also discussed in light of the nature of external/internal and nondefect/defect sites of TS-1.