OBJECTIVE: This study was undertaken to determine which factors were statistically predictive of radiological, and clinical outcomes. in the radiosurgical. treatment-of arteriovenous malformations (AVMs). METHODS: The computerized dosimetric and clinical data for 269 patients were reviewed. The AVM nidus was hand-contoured on successive enhanced computed tomographic slices through the nidus, to allow detailed determinations of nidus volume, target miss, normal brain, tissue treated, dose conformality, and dose gradient. In addition, a number of patient and treatment factors, including Spetzler-Martin grade, presenting symptoms, dose, number of isocenters, radiological outcome, and clinical outcome, were subjected to multivariate analysis. RESULTS: Two hundred twenty-five patients were treated with radiosurgery-for the first time, and 44 patients underwent radiosurgical retreatment. One hundred forty-three-patients had AVMs located in or near "eloquent" brain areas and 126 patients did not; Seventy patients demonstrated preoperative neurological findings related to the AVM and 199, did not. Twenty-six patients had previously undergone endovascular treatment and 10 patients had previously undergone surgical treatment of their AVMs. Of the 269 patients studied, 228 experienced no complication, 10 (3.7%) experienced a transient radiation-induced complication, 3 (1 %) experienced a permanent radiation-induced complication, and 28 (10%) experienced posttreatment hemorrhage. CONCLUSION: None of the analyzed factors was predictive of hemorrhage after radio-surgery in this study. The 12-Gy volume was predictive of permarientracli complications. Eloquent AVM location and 12-Gy volume were correlated with the occurrence of transient radiation-induced complications. Better conformality was correlated with a reduced incidence of transient complications. Lower Spetzler-Martin grades, higher doses, and steeper dose gradients were correlated with radiological success.