Accurate knowledge of cerebral aneurysm volume would be valuable in guiding the volume of embolized material required for optimal filling of an aneurysm sac and recording percentage volume filling. Algebraic volumes are frequently estimated by algebraic volume formulae. 3D digital subtraction angiography (DSA) aids endovascular treatment planning and yields volumetric data. Our aim was to define the accuracy of 3D-DSA in quantifying aneurysm volume using an automated voxel-based volumetric method (voxel volume method) and compare results to volumes calculated by ellipsoid and cylindrical algebraic formulae (algebraic volume method). We constructed 13 latex aneurysm moulds and measured their true volumes using a micropipette in-vitro. 3D-DSA was performed on contrast filled moulds and experimental volume estimated by both voxel and algebraic methods. In our in-vivo study we quantified the voxel and algebraic volumes from the 3D data sets of 75 cerebral aneurysms. The linear regression test provided correction values between voxel and algebraic methods. The in-vitro study showed that the voxel volume method was the most accurate (mean percentage deviation from true volume 3.7 +/- 3.5%; p=0.9). The ellipsoid method significantly underestimated - 11.2 +/- 13.6%; p < 0.05) and the cylindrical method overestimated (42.6 +/- 35.7%; p < 0.05) true aneurysm volume. Similar results were obtained in-vivo. While algebraic measurements could be corrected by an equation, the clinical usefulness of this equation is questionable due to the large volume range to achieve a 95% confidence interval. The voxel volume method is accurate in quantifying aneurysm volume. Aneurysms in-vivo do not conform to simple algebraic geometry. Aneurysm volume on 3D-DSA should be calculated by the voxel-based method and not by algebraic formulae.