The rationale for study of proton radiation therapy is that, for some anatomic sites and tumors, the treatment volume is smaller; i.e., there is less irradiation of nontarget tissue while the target is included in three dimensions at each treatment session. As a result, the dose to the target can be raised. The consequence is that the tumor control probability improves and the frequency and severity of treatment-related morbidity decrease. These results come about from the physical fact that the proton range in tissue is finite; in comparison, absorption of photons is an exponential function and, hence, some dose is received for the full-beam path through the body. Accordingly, the dose deep to the target for proton treatments can be zero for each beam path. This situation provides a virtually certain means of improving the treatment outcome for selected categories of patients. Experience to date with proton radiation therapy has been quite limited. As of June 1991, the total number of proton radiation-treated patients was 11,763 from the various centers. Of that number, approximately 46% and 32% have been treated for small benign intracranial lesions (principally pituitary adenomas and arteriovenous malformations) and for tumors of the eye, respectively. Thus, only some 2500 patients have been treated for all other tumor types. The results from three centers and approximately 2800 patients with uveal melanoma are that the local control rate was 96% (for failures in-field, marginal, and in other parts of the eye). The local control results for chondrosarcomas and chordomas of the skull base are 91% and 65%, respectively. These percentages compare with some 35% achieved with conventional treatment. Experience with arteriovenous malformations indicates that control of bleeding and disappearance of the lesion are comparable to those achieved by other procedures. The developments from the proton therapy programs have contributed greatly to radiation treatment planning, e.g., the first three-dimensional treatment planning system put into regular clinical use (uveal melanoma), beam's eye view, digital-reconstructed radiograph, dose-volume histograms, and definitions of the uncertainty in dose around any defined point. The potential for clinical gains is high. In May 1991, the Proton Radiation Oncology Group was formed to design, supervise, and coordinate clinical trials and to assist in data analysis. The efficacy of proton radiation therapy will be compared with that of photon therapy of the very highest technology.