PURPOSE: To determine, by means of a phantom study, the distortion-related factors and appropriate iodine concentration for three-dimensional reconstruction rotational angiography. MATERIALS AND METHODS: Four phantoms were created: crossed metal rods, one metal rod, one contrast rod, and a contrast rod under water. Iodine concentrations were 300, 250, 200, and 150 Img/ml, respectively. For each phantom, rotational angiography was performed in the rotational, right-angled (90 degree to rotational), intermedial (45 degree), close to rotational (20 degree), and close to right-angled (70 degree) planes. Two-dimensional projection images were transferred to a workstation at which 3D images were produced using the volume rendering technique. Image quality in each plane was evaluated in terms of opacity, homogeneity, and margin sharpness, which were graded as low, intermediate or high by two neuroradiologists who used images obtained in the right-angled plane as the standard reference. The same assessors evaluated in terms of the same parameters, cross-sectional images obtained at the central, intermedial, and peripheral portions of one metal rod positioned in the right-angled, close to right-angled, and intermedial planes, and in order to compare the values at different sites, one neuroradiologist measured the horizontal and vertical diameters of each cut image. RESULTS: Three-dimensional images of all four phantoms were high quality in the close to right-angled and intermedial plane, but in the rotational and close to rotational plane were degraded. In particular, metal rod images obtained in the rotational plane were poor for all three items. In these two planes, image quality was better for the contrast rod than the metal rod, and at 200 and 250 Img/ml concentrations than at 300 and 150 Img/ml concentrations. There was no significant difference in image quality, nor in measured values of the diameter between cut images. CONCLUSION: A three-dimensional image was more distorted when a linear object was placed at a lesser angle to the rotational plane and when inherent X-ray attenuation was greater, a finding which must be closely related to the beam-hardening artifact. Distortion was least at 200-250 Img/ml of iodine concentration, the concentration thought to be most appropriate for in-vitro 3D angiography.