Sixteen cerebral arteriovenous malformations (AVMs) were examined to determine the role of magnetic resonance (MR) imaging in verifying obliteration. The AVMs (mean volume 7.5 cm3, range 2-17 cm3) were treated with Gamma Knife surgery between March 1993 and May 1994. Integration of stereotactic MR and stereotactic conventional X-ray angiography (XRA) was used for targeting in the Gamma Knife surgery. All MR examinations both for targeting and follow-up, were performed on a 1.5-Tesla superconductive MR scanner (Signa). Multiple pulse sequences (spin echo T1- and T2-weighted MR imaging, and three-dimensional time-of-flight MR angiography) were used. The mean maximum target dose was 37.4 Gy (range 25.0-44.0 Gy). The mean minimum target dose was 20.1 Gy (range 17.5-25.2 Gy). Follow-up imaging was performed about every 6 months or when clinically warranted. XRA was performed when the AVM was no longer seen on MR images. The time from the last MR image to the XRA was within 1 month in nine patients, and longer in seven. MR imaging demonstrated regressing AVMs in all patients as early as 3 months after Gamma Knife surgery. For seven of the nine patients total obliteration on MR was confirmed on XRA within 1 month. In the other two, previous hemorrhage and adverse radiation effects probably caused overestimation of AVM obliteration. In the remaining seven patients, XRA confirmed the MR observation although the time intervals were longer. It is concluded that, for medium- to large-volume AVMs, MR can demonstrate not only the regressing AVMs but also verify total obliteration. However, verification has to be based on an integration of MR imgaging and MR angiography. The use of MR reduces the invasiveness of Gamma Knife surgery for cerebral AVMs.