Twenty-one arteriovenous malformations were prospectively evaluated using magnetic resonance angiography and compared with stereotactic angiography. The goals were to establish the feasibility of magnetic resonance angiography, compare it to stereotactic angiography, employ magnetic resonance angiography in follow-up, and semiquantify flow. A correlative evaluation between flow and response to stereotactic radiosurgery was carried out. Phase contrast angiograms were obtained at flow velocities of 400, 200, 100, 60, and 20 cm/sec. The fractionated velocities provided images that selectively demonstrated the arterial and venous components of the arteriovenous malformations. Qualitative assessment of the velocity within the arteriovenous malformations and the presence of fistulae were also determined by multiple velocity images. In addition, 3-dimensional time-of-flight magnetic resonance angiograms were obtained to define the exact size and shape of the nidus. This technique also permitted evaluation of the nidus and feeding arteries for the presence of low flow aneurysms. Correlation between the two imaging modalities was carried out by subjective and semiquantitative estimation of flow velocity and estimation of nidus size. The following velocity parameters were employed: fast, intermediate, slow, and none (arteriovenous malformation obliterated). In 19 of 21 (90.5%) arteriovenous malformations, magnetic resonance angiography was equal or superior to stereotactic angiography for flow quantification and visualization of the nidus. Only 2 of 21 arteriovenous malformations were better demonstrated by stereotactic angiography than by magnetic resonance angiography (failure rate of 9.5%). The nidus size in one case was clearly underestimated by stereotactic angiography and would have resulted in a geographic miss without magnetic resonance angiography. Seven post-radiosurgery arteriovenous malformations were evaluated for follow-up with both magnetic resonance angiography and stereotactic angiography. In 6 of 7 arteriovenous malformations, magnetic resonance angiography response matched stereotactic angiography response. Correlation of flow with outcome was carried out for 14 arteriovenous malformations using magnetic resonance angiography only. Interestingly, all nine arteriovenous malformations with intermediate or slow flow demonstrated partial or complete obliteration; whereas only 3 of 5 fast flow arteriovenous malformations achieved a response with a median follow-up of 10 months. This early analysis suggests that slower flowing arteriovenous malformations may obliterate faster after stereotactic radiosurgery and flow parameters could be employed to predict response. In conclusion, magnetic resonance angiography permits semiquantitative flow velocity assessment and may therefore be superior to stereotactic angiography. An additional advantage of magnetic resonance angiography is the generation of serial transverse images which can replace the conventional CT scan employed for stereotactic radiosurgery treatment planning.(ABSTRACT TRUNCATED AT 400 WORDS)