Some 9000 patients throughout the world have been treated by some form of neutron beam therapy. These include patients with advanced nonresectable tumors in many different sites treated with a variety of neutron beam generators varying widely in beam energy. Protocols were largely nonrandomized and included both mixed beam studies (neutrons + photons) and neutrons alone in varying doses. In spite of wide variation in equipment, treatment technique, and philosophy, some consistent trends have been identified: (1) in general, the neutron results have been at least as good as those of the photon controls measured in terms of local control, although the incidence of significant side effects have been higher; (2) in none of the randomized studies conducted so far, largely comprising epidermoid carcinomas of the head and neck, has a clear survival advantage for neutrons over photon controls been demonstrated at a statistically significant level; (3) results with mixed beam studies have been uniformly equivocal, with marginally significant differences in favor of the experimental groups compared with the photon controls; (4) adenocarcinomas of the gastrointestinal tract (GI) tract, including tumors of the salivary gland, pancreas, stomach, and bowel, appear to be responsive to high linear energy transfer (LET) radiation; (5) nonepidermoid, radioresistant tumors (sarcoma of bone and soft tissue and melanoma) yield a consistantly high local control rate, with neutron irradiation strikingly superior to those reported with photon therapy; and (6) in the central nervous system, both normal tissues and tumors appear to be exceptionally sensitive to neutron irradiation, therapeutic ratios are small, and the prospect of cure remains remote. It is concluded that neutrons are efficacious for certain specific tumor types, but that essentially new study designs, based on nonrandomized matched case comparisons, will be required to prove the merit of the new modality.