Until recently, the inability to demonstrate neurovascular compression of the trigeminal nerve preoperatively resulted in surgery being offered only in cases of severe trigeminal neuralgia (TGN), frequently after a prolonged trial of medical treatment and following less invasive procedures, despite the fact that posterior fossa microvascular decompression gives long-term pain relief in 80% to 90% of cases. To assess whether vascular compression of the nerve could be demonstrated preoperatively, high definition magnetic resonance tomographic angiography (MRTA) was performed in 50 consecutive patients, five of whom had bilateral TGN, prior to posterior fossa surgery. The imaging results were compared with the operative findings in all patients, including two patients who underwent bilateral exploration. Vascular compression of the trigeminal nerve was identified in 42 of 45 patients with unilateral symptoms and on both sides in four patients with bilateral TGN. In the last patient with bilateral TGN, neurovascular compression was identified on one side, and on the other side the compressing superior cerebellar artery was separated from the nerve by a sponge placed during previous surgery. There was full agreement regarding the presence or absence of neurovascular compression demonstrated by MRTA in 50 of 52 explorations, but MRTA misclassified four vessels compressing the trigeminal nerve as arteries rather than veins. In two cases, there was disagreement between the surgical and MRTA findings. In the first of these cases, surgery revealed distortion of the nerve at the pons by a vein that MRTA had predicted to lie 6 mm remote from this point. In the second patient, venous compression was missed; however, this patient was investigated early in the series and did not have gadolinium-enhanced imaging. In nine cases, MRTA correctly identified neurovascular compression of the trigeminal nerve by two arteries. Moreover, MRTA successfully guided surgical reexploration in one patient in whom a compressing vessel was missed during earlier surgery and also prompted exploration of the posterior fossa in two patients with multiple sclerosis and one patient with Charcot-Marie-Tooth syndrome, in whom neurovascular compression was identified preoperatively. It is concluded that MRTA is an extremely sensitive and specific method for demonstrating vascular compression in TGN. As a result, open surgical procedures can be recommended with confidence, and microvascular decompression is now the treatment of choice for TGN at the authors' unit. They propose MRTA as the definitive investigation in such patients in whom surgery is contemplated.