High confining pressure fracture tests of Indiana limestone <Abou-Sayed, 1977> and Iidate granite were simulated using boundary element techniques and a Dugdale-Barenblatt (tension-softening) model of the fracture process zone. Our results suggest a substantial (more than a factor of 2) increase in the fracture energy of Indiana limestone when the confining pressure was increased from zero to only 6--7 MPa. While Hashida et al. <1993> concluded that there was no change in the fracture energy of Iidate granite at confining pressures up to 26.5 MPa, we find that data from one series of experiments (compact-tension tests in their terminology) are also consistent with a significant (more than a factor of 2) increase in fracture energy. Data from another set of their experiments (thick-walled cylinder tests) seem to indicate a decrease in the fracture energy of Iidate granite at confining pressures of 6--8 MPa, but these may be biased due to the very small specimen size. To our knowledge these results are the first reliable indication from laboratory experiments that rock tensile fracture energy varies with confining pressure. Based on these results, some possible mechanisms of pressure sensitive fracture are discussed. We suggest that the inferred increase in fracture energy results from more extensive inelastic deformation near the crack tip that increases the effective critical crack opening displacement. Such deformation might have occurred due to the large deviatoric stress in the vicinity of the crack tip in the Abou-Sayed experiments, and due to the enlarged region of significant tensile stress near the crack tip in the Hashida et al. compact tension tests. These results also highlight the fact that at confining pressures that exceed the tensile strength of the material, tensile fracture energy will in general depend upon the crack size and the distribution of loads within it, as well as the ambient stress.¿ 1997 American Geophysical Union