Neutrons are by-products of high-energy radiation therapy and a source of dose to normal tissues. Thus, the presence of neutrons increases a patient's risk of radiation-induced secondary cancer. Although neutrons have been thoroughly studied in air, little research has been focused on neutrons at depths in the patient where radiosensitive structures may exist, resulting in wide variations in neutron dose equivalents between studies. In this study, we characterized properties of neutrons produced during high-energy radiation therapy as a function of their depth in tissue and for different field sizes and different source-to-surface distances (SSD). We used a previously developed Monte Carlo model of an accelerator operated at 18 MV to calculate the neutron fluences, energy spectra, quality factors, and dose equivalents in air and in tissue at depths ranging from 0.1 to 25 cm. In conjunction with the sharply decreasing dose equivalent with increased depth in tissue, the authors found that the neutron energy spectrum changed drastically as a function of depth in tissue. The neutron fluence decreased gradually as the depth increased, while the average neutron energy decreased sharply with increasing depth until a depth of approximately 7.5 cm in tissue, after which it remained nearly constant. There was minimal variation in the quality factor as a function of depth. At a given depth in tissue, the neutron dose equivalent increased slightly with increasing field size and decreasing SSD; however, the percentage depth-dose equivalent curve remained constant outside the primary photon field. Because the neutron dose equivalent, fluence, and energy spectrum changed substantially with depth in tissue, we concluded that when the neutron dose equivalent is being determined at a depth within a patient, the spectrum and quality factor used should be appropriate for depth rather than for in-air conditions. Alternately, an appropriate percent depth-dose equivalent curve should be used to correct the dose equivalent at the patient surface.