The pathology of malignant brain tumors often precludes successful treatment by surgery and standard radiation therapy. Boron neutron-capture therapy consists of the selective loading of tumor with 10B and subsequent irradiation with a thermal or epithermal neutron field. The neutron-capture reaction 10B(n,α)7Li produces high-linear-energy-transfer-charged particles that deposit energy principally within the abnormal tissue that contains a high 10B concentration. Constraints on this therapy modality are imposed by radiation effects in normal tissue from thermal neutrons, neutron-induced gamma rays, fast-neutron and gamma-ray beam contaminants, and also from the 10B(n,α)7Li reactions in circulating blood. The ANDY general geometry Monte Carlo code is used to calculate the space-energy distribution of all pertinent components of the dose within a simple head phantom in an idealized therapy configuration at the Massachusetts Institute of Technology Research Reactor. The effects of 10B concentration, gamma-ray contamination of the therapy beam, thermal neutron beam aperture, and surgically formed re-entrant cavities are examined with respect to several clinical criteria for therapeutic efficacy. It is found for the model considered that the maximum effective relaxation length for the thermal neutron fluence is 1.6 to 1.8 cm, which is 30 to 40% lower than the infinite medium relaxation length, and thereby indicates the importance of multidimensional boundary effects in this calculation. The fluence-depth characteristic was verified by an experimental irradiation of a tis-sue-equivalent head phantom with re-entrant cavityy and excellent agreement was observed between measured and calculated results. It was also found that the gamma-ray beam contaminant is not necessarily deleterious to therapeutic efficacy, that a larger aperture thermal neutron beam improves the dose field with respect to some criteria but at the expense of others, and that plausible size variations in the surgically formed cavity do not change the character of the dose field. As a further refinementy a Monte Carlo microdosimetry model is developed and applied to the problem of radiation effects on the cerebral microvasculature by 10B capture reactions in the circulating blood. Qualitative predictions of this model correlate positively with previous clinical experience.