Experiments were performed to quantitatively confirm the relationship between open dry microcracks and P wave velocity on crystalline rocks. Equations derived by Hudson <1980> (first-order corrections only) describe the experimentally obtained relationship. Microcrack analysis consists of chemical contact imagery to make the cracks visible, image analysis, and stereologic data processing to obtain three-dimensional crack characteristics from data obtained in three orthogonal plane images. Pressure-dependent measured P wave velocities are used to estimate intrinsic P wave velocities (velocities characteristic for the uncracked rock) by extrapolating the linear part of the pressure-velocity curve from high pressure to low pressure. Therefore the impact of open microcracks on P wave velocity is given by normalized P wave velocities, which are the ratio between measured and intrinsic velocities at a hydrostatic pressure of 1 MPa. Three-dimensional crack characteristics and normalized P wave velocities, are obtained in three orthogonal directions. Velocities in all other directions were estimated by interpolation assuming an orientation distribution that can be described by a triaxial ellipsoid. In this way it is possible to study the relationship between open dry cracks and P wave velocity experimentally and to compare the results with the theoretically predicted relationship using one single model. In a case study, pressure-dependent measured P wave velocities have been used to infer the distribution of open microcracks in 84 drill cores from the Continental Deep Drilling Project of the Federal Republic of Germany (Kontinentales Tiefbohrprogramm der Bundesrepublik Deutschland, or KTB) covering the depth range between 621-8080 m. The cores were studied after stress relaxation (1--7 years after drilling). It was not possible to obtain information about the in situ stress field owing to the anisotropic nature of the rock fabric. ¿ 1999 American Geophysical Union