To determine whether the catalytic roles of extracrystalline and intracrystalline gold nanoparticles supported on titanosilicate-1 (TS-1) for direct propylene epoxidation are intrinsically different, the kinetics of direct propylene epoxidation were measured in a gas-phase continuous stirred tank reactor (CSTR) over polyvinylpyrrolidone-coated (PVP) gold nanoparticles (Au-PVP/TS-1) deposited on TS-1. The as-made PVP-coated gold nanoparticles were too large to fit into the micropores of TS-1, even after ligands were removed in situ by a series of pretreatments, as confirmed by both TEM and TGA-DSC. The activation energy (51 kJ mol–1) and reaction orders for H2 (1.3), O2 (0.4), propylene (0.4), propylene oxide (−0.6), carbon dioxide (0), and water (0) for propylene epoxidation measured on Au-PVP/TS-1 were consistent with those reported for Au/TS-1 prepared via deposition-precipitation (Au-DP/TS-1) (52 kJ mol–1, H2, 1; O2, 0.4; C3H6, 0.4; C3H6O, −0.6; CO2, 0; H2O, 0). However, while the reaction orders for hydrogen oxidation on Au-PVP/TS-1 (H2, 0.8; O2, 0; C3H6, −0.3; C3H6O, −0.1; CO2, 0; H2O, −0.2) were similar to those measured on Au-DP/TS-1 (H2, 0.9; O2, 0.3; C3H6, −0.3; C3H6O, 0; CO2, 0; H2O, −0.1), a decrease in activation energy from approximately 30 kJ mol–1 for Au-DP/TS-1 to 4–5 kJ mol–1 for Au-PVP/TS-1 suggests there is a change in rate-limiting step or active site for hydrogen oxidation. Additionally, an active site model was developed which determines the number of Ti within an interaction range of the perimeter of extracrystalline Au nanoparticles (i.e., the number of Au–Ti active site pairs). This model was used to estimate catalytic turnover frequencies over solely proximal Au–Ti pairs, assuming that hydrogen peroxide does not desorb and migrate from Au sites to Ti sites and instead propylene oxide forms in a concerted mechanism previously termed the “simultaneous” mechanism. Turnover frequencies estimated for this active site model for a data set containing both Au-DP/TS-1 and Au-PVP/TS-1 were ∼20× higher than the highest previous reported estimates (∼80 s–1 vs 1–5 s–1 at 473 K) for catalytic oxidation on noble metals, suggesting that the simultaneous mechanism occurring over proximal Au–Ti sites alone is incapable of explaining the observed rate of propylene epoxidation and that short-range migration of hydrogen peroxide is kinetically relevant. The agreement of reaction orders, activation energy, and active site model for propylene epoxidation on both Au-DP/TS-1 and Au-PVP/TS-1 suggests a common mechanism for propylene epoxidation on both catalysts containing small intraporous gold clusters and catalysts with exclusively larger extracrystalline gold nanoparticles. Rates of hydrogen oxidation were found to vary proportionally to the amount of surface gold atoms. This is also consistent with the hypothesis that the observed decrease in hydrogen efficiency and PO site-time-yield per gold mass with increasing gold loading are driven primarily by the gold dispersion in Au/TS-1 catalysts.