The hypothesis of boron neutron capture therapy (BNCT) research has been that the short-range, high-linear energy transfer radiation produced by the capture of thermal neutrons by (10)B will potentially control tumor and spare normal tissue only if the boron compound selectively targets tumor tissue within the treatment volume. In a previous in vivo study of low-dose BNCT mediated by GB-10 (Na(2)(10)B(10)H(10)) alone or combined with boronophenylalanine (BPA) in the hamster cheek pouch oral cancer model that was primarily designed to evaluate safety and feasibility, we showed therapeutic effects but no associated normal tissue radiotoxicity. In the present study, we evaluated the response of tumor, precancerous and normal tissue to high-dose BNCT mediated by GB-10 alone or combined with BPA. Despite the fact that GB-10 does not target hamster cheek pouch tumors selectively, GB-10-BNCT induced a 70% overall tumor response with no damage to normal tissue. (GB-10+BPA)-BNCT induced a 93% overall tumor response with no normal tissue radiotoxicity. Light microscope analysis showed that GB-10-BNCT selectively damages tumor blood vessels, sparing precancerous and normal tissue vessels. In this case, selective tumor lethality would thus result from selective blood vessel damage rather than from selective uptake of the boron compound.