The rate of hydrogen diffusion in clinopyroxene is relevant to interpreting hydrogen (“water”) concentrations in xenoliths, phenocrysts, and clinopyroxene-hosted melt inclusions to provide insight into the deep-earth water cycle and volcanic explosivity. Here, we determine bulk and site-specific hydrogen diffusivities in two diopsides and an augite by heating initially homogeneous water-bearing samples in a 1-atm CO/CO2 gas-mixing furnace at 800–1000 °C and oxygen fugacity at the quartz–fayalite–magnetite buffer and observing H-loss profiles. The O–H stretching range between wavenumbers 3000 and 4000 cm−1 in FTIR spectra is resolved into 4–6 peaks, each of which is assumed to represent a distinct defect site for the hydrogen, to determine peak-specific diffusivities using our previously published whole-block method. For the diopside from the Kunlun Mts. in China, Arrhenius relations are reported for peaks at 3645, 3617, 3540, 3443, and 3355 cm−1 based on measurements at 816, 904, and 1000 °C. Bulk and site-specific diffusivities are determined for the same set of peaks at 904 °C for the second diopside (Jaipur). The augite (PMR-53) was a triangular thin slab, and hydrogen diffusivities were determined for bulk hydrogen and peaks at 3620, 3550, 3460, and 3355 cm−1 in the thickness direction at 800 °C. Bulk hydrogen diffusivity in the Jaipur diopside is consistent with previous work, and hydrogen diffusivity in augite PMR-53 is slightly lower than the fast direction diffusivities measured ||  and * in Jaipur diopside. Both diopsides show 1–2 orders of magnitude differences in the peaks-specific diffusivities, with the fastest diffusivities at 3450 cm−1 and the slowest at 3645 cm−1. However, the hydrogen diffusivities in Jaipur diopside are 2–4 orders of magnitude higher than those in Kunlun diopside for bulk hydrogen and all peaks. Thus, peak-specific differences cannot by themselves adequately explain the 5 orders of magnitude range in hydrogen diffusivities observed experimentally in different diopsides. The results are broadly consistent with a previously proposed increase in hydrogen diffusivity in diopside with Fe up to 0.6–0.8 a.p.f.u., although there may be an opposing relationship with Al(IV). The results of this study and others predict high water diffusivities in Fe-bearing mantle xenolith clinopyroxene, on the order of ~10−9 to 10−11 m2/s at 1000 °C. The common observation of hydrogen zonation in mantle xenolith olivine, but not in clinopyroxene implies that hydrogen diffusion is much faster in olivine than in pyroxene, which then requires the operation of the fastest diffusion mechanism quantified in olivine and diffusivities in clinopyroxene at the lower end of this 2 orders of magnitude range. Such high diffusivities strongly suggest that water in mantle xenoliths has at least partially equilibrated with the host magma, and that the diffusion profiles observed in mantle xenolith olivine reflect only the final stage of ascent after water begins to exsolve.