We set out to validate the concept of three-dimensional (3D) angiography. We evaluated the sensitivity and the quality of morphological analysis mode possible by an experimental system for imaging cerebrovascular disease versus standard digital subtraction angiography (DSA). The system, the 3D Morphometer, is a computerised X-ray angiography unit capable of acquiring a set of two-dimensional (2D) projections during a rotation and then reconstructing a 3D volume from them. We studied 78 patients with suspected cerebrovascular disease. 3D and 2D images (standard 2D DSA performed during the same procedure), were reviewed blindly to assess detection and display of morphological characteristics of cerebrovascular diseases. We found 53 aneurysms, 22 arteriovenous malformations and two venous angiomas. On 3D angiography we detected two aneurysms we missed on 2D angiography. In 47 aneurysms on which further data were obtained during surgery or embolisation, the 3D angiography allowed more accurate analysis of the neck and surrounding vessels in cases in which the 2D angiographic findings were doubtful. Assessment of arteriovenous malformations was equivalent with both techniques. Under the conditions of our study, the technical constraints being the same for both methods, 3D angiography was superior to 2D angiography. Implementation on C-arm vascular systems is being evaluated.