Hydrogen is a widespread trace element in many nominally anhydrous minerals (NAMs) from the Earth's crust and mantle. Hydrogen is normally incorporated in the form of hydroxyl ions and can be regarded as structurally bound water. The most important minerals of the upper mantle: olivine, orthopyroxene, clinopyroxene and garnet, all contain small but significant amounts of hydrogen, and the upper mantle has a capacity to store the equivalent of several world oceans. The main objective of this project is to investigate in detail the mechanisms that are responsible for water incorporation in upper mantle pyroxenes (and synthetic analogues) as pyroxenes can carry substantial amounts of water, up to 1300 wt ppm H2O. Fundamental questions are how much of the original xenolith water that is lost during transport to the surface and if the spectroscopic features measured in the minerals are representative for mantle conditions. The redox reaction: Fe2+ + OH- ↔ Fe3+ + O2- + ½H2, which is relatively fast, is thought to be the dominant hydrogen exchange reaction in many minerals (Ingrin & Skogby, 2000). The reaction is fast enough to suggest that water in NAMs equilibrates with the transporting magma and related fluids during ascent to the surface. Nevertheless, several studies show systematic variations in water content with geological environment (Bell & Rossman, 1992; Peslier et al., 2002), implying a complex relationship between host mineral, mantle source region, magma type and eruption style. The current part of the project is focused on the dehydration-hydration mechanisms in diopside, the most common variety of clinopyroxene in the upper mantle. The approach has been to study the kinetics and temperature dependence of the reactions controlling hydrogen diffusion in synthetic Fe-poor diopside. Other reactions are likely to be obscured by the iron redox reaction if measured in natural mantle diopside containing several wt% FeO. Therefore, synthetic diopside with very low amounts (0.7 wt% FeO) of iron had to be used in order to measure the influence and co-dependence of the iron redox reaction with other possible reactions.