Seismic investigations using three-component instruments have demonstrated shear wave splitting in widely different geological and tectonic environment. Shear wave analysis suggests that the inferred anisotropy can best be explained if the rocks contain distributions of subparallel, subvertical, fluid-filled microfractures which are aligned according to the current stress field (the Crampin ''extensive dilatancy anisotropy'' (EDA) model). Geological evidence for widespread fluid activity in the Earth's crust supports this hypothesis and suggests that the myriads of planes of microscopic secondary fluid inclusions which characterise most geological materials represent fossil EDA cracks. Field studies confirm a strong spatial relationship between the orientation of fluid inclusion arrays and the paleodirections of maximum compressive stress. At elevated temperatures and pressures, the healing of fluid-filled cracks in the crust, of whatever origin, is considered to be a fairly rapid process. Moreover, the inclusions which form as a result of crack healing respond to changes in temperature, pressure, and stress by changing shape and/or volume. This indicates a high degree of fracture compliance; sufficient perhaps to react to rapid changes in the current stress field and associated shear wave splitting. Though individual inclusions (typically <20 μm) will have no effect on the propagation of seismic waves with wavelengths of tens to hundreds of meters, the rock volume sampled by the waves will be rendered anisotropic if the overall average of the inclusions shows a preferred orientation. Thus fluid inclusions may constitute one of the principal types of fluid-filled microcrack described by the EDA hypothesis. ¿ American Geophysical Union 1990