Between May, 1988 and August, 1993, 158 patients with arteriovenous malformations (AVMs) were treated radiosurgically at the University of Florida. A mean dose of 1560 cGy was directed to the periphery of the lesions, which had a mean volume of 9 cc (0.5 to 45.3 cc). One hundred thirty-nine of these individuals were treated with one isocenter. The mean follow-up interval was 33 months with clinical information available on 153 of these patients. Patients were followed until magnetic resonance (MR) studies suggested complete AVM thrombosis. An arteriogram was then performed, if possible, to verify occlusion status. If arteriography revealed any persistent nidus at 36 months posttreatment, the residual nidus was re-treated. Outcome categories of AVMs analyzed included the following possibilities: 1) angiographic cure; 2) angiographic failure; 3) re-treatment; 4) MR image suggested cure; 5) MR image suggested failure; 6) patient refused follow-up evaluation; 7) patient lost to follow-up study; or 8) patient deceased. The endpoints for success or failure of radiosurgery were as follows: angiographic occlusion (success), re-treatment (failure), and death due to AVM hemorrhage (failure). Fifty-six patients in this series reached one of the endpoints. Successful endpoints were seen in 91% of AVMs between 1 and 4 cc in volume, 100% of AVMs 4 to 10 cc in volume, and 79% of AVMs greater than 10 cc in volume. The more traditional measure of radiosurgical success, percentage of angiograms showing complete obliteration, was obtained in 81% of AVMs between 1 and 4 cc in volume, 89% of AVMs between 4 and 10 cc in volume, and 69% of AVMs greater than 10 cc in volume. A detailed analysis of the relationship of all outcome categories to size is presented.