We report a B3LYP study of a novel mechanism for propylene epoxidation using H-2 and O-2 on a neutral Au-3 cluster, including full thermodynamics and pre-exponential factors. A side-on O-2, adsorption on Au-3 is followed by dissociative addition of H-2 across one of the Au-O bonds (Delta E-act = 2.2 kcal/mol), forming a hydroperoxy intermediate (OOH) and a lone H atom situated on the Au-3 cluster. The more electrophilic O atom (proximal to the An) of the Au-OOH group attacks the C C of an approaching propylene to form propylene oxide (PO) with an activation barrier of 19.6 kcal/mol. We predict the PO desorption energy from the Au-3 cluster with residual OH and H to be 11.5 kcal/mol. The catalytic cycle can be closed in two different ways. In the first subpathway, OH and H, hosted by the same terminal Au atom, combine to form water (Delta E-act = 26.5 kcal/mol). We attribute rather a high activation barrier of this step to the breaking of the partial bond between the H atom and the central Au atom in the transition state. Upon water desorption (Delta E-des = 9.9 kcal/mol), the Au-3 is regenerated (closure). In the second subpathway, H-2 is added across the Au-OH bond to form water and another Au-H bond (Delta E-act = 22.6 kcal/mol). Water spontaneously desorbs to form an obtuse angle Au-3 dihydride, with one H atom on the terminal An atom and the other bridging the same terminal An atom and the central Au atom. A slightly activated rearrangement to a symmetric triangular Au-3 intermediate with two equivalent Au-H bonds, addition of 0, into the Au-H bond, and rotation reforms the hydroperoxy intermediate in the main cycle. On the basis of the Delta G(act), which contains contribution from both pre-exponetial factor and activation energy, we identify the propylene epoxidation step as the actual rate-determining step (RDS) in both the pathways. The activation barrier of the RDS (epoxidation step: Delta E-act = 19.6 kcal/mol) is in the same range as that in the published computationally investigated olefin epoxidation mechanisms involving Ti sites (without Au involved) indicating that isolated Au clusters and possibly Au clusters on non-Ti supports can be active for gas-phase partial oxidation, even though cooperative mechanisms involving Au clusters/ Ti-based-supports may be favored.