The purposes of this study were to determine the safety and efficacy of embolization of brain AVMs prior to radiosurgery and to evaluate the total obliteration rate achieved. The brain AVMs of 64 patients were subselectively embolized mainly with NBCA, platinum microcoils and/or PVA. The aim of embolization was the reduction of the target volume and/or the elimination of vascular structures bearing an increased risk of hemorrhage. Presenting symptoms were intracranial hemorrhage in 33 patients, a seizure disorder in 21 patients, and headache in 6 patients. Four AVMs had been detected as an incidental finding. The initial AVM volume was in the range of 0.5 to 84 cc (mean 17 cc). Grading of the AVMs according to the Spetzler-Martin scale showed the following distribution: grade I, 3x; II, 13x; III, 11x; IV, 17x; V, 4x; VI, 16x. A total of 300 endovascular procedures including 47 subselective catheterizations without and 253 with embolization were performed. A size reduction of the AVMs between 10% and 95% (mean 63%, median 70%) was achieved. Neurological symptoms due to treatment complications were transient in 12 patients, of minor clinical significance but permanent in 4 patients. Following radiosurgery, one patient died due to recurrent intracerebral hemorrhage. Three patients are doing well but refused final follow-up angiography. A total of 30 patients is currently within the latency interval after radiosurgery. Radiosurgery failed to obliterate the embolized AVMs in 16 patients. Angiography confirmed complete nidus obliteration in 14 patients. The endovascular treatment of brain AVMs prior to radiosurgery proved safe and effective and may be considered in either high grade or incidental AVMs. AVM obliteration after embolization and radiosurgery is less frequently achieved than after stereotactic irradiation of primarily small AVMs.