Methods of stereotactic radiosurgery are reviewed and compared with respect to technical factors and published clinical results. Heavy-ion beams, the Leksell cobalt-60 gamma knife, and the conventional linear accelerator (linac) are compared with respect to dosimetry, radiobiology, treatment planning, cost, staffing requirements, and ease of use. Clinical results on the efficacy of treatment of arteriovenous malformations are tabulated, and other applications of radiosurgery are described. It is concluded that although there are dosimetric and radiobiological advantages to charged-particle beams that may ultimately prove critical in the application of radiosurgery to large (> 30 mm) lesions, these advantages have not yet demonstrated clinical effect. On the other hand, equally excellent clinical results are obtained for small lesions with photon beams--the gamma knife and the linac. There are only minor differences between gamma and x-ray beam dose distributions for small, spherical-shaped targets. Mechanical precision is superior for the gamma knife as compared with the linac. The superior mechanical precision is of limited importance for most clinical targets, because inaccuracy of cranial target localization based on radiological imaging is greater than the typical linac imprecision of +/- 1 mm. Treatment planning for the linac is not standardized, but existing systems are based on well-known algorithms. The linac allows flexible, ready access to individualized beam control, without intrinsic field size limitations. Thus, it is more readily possible to achieve homogeneous dose distributions for nonspherical targets with one or more dimensions greater than 25 mm, as compared with that achieved with the gamma unit.(ABSTRACT TRUNCATED AT 250 WORDS)