OBJECTIVE: This study aimed at comparing three-dimensional (3-D) reconstruction with two-dimensional coronary angiograms with respect to anatomical parameters that might affect plaque formation and rupture. METHODS: Sixty patients with stable left anterior descending (LAD) lesions and 60 patients with an anteroseptal myocardial infarction and recanalized LAD were studied. RESULTS: Conventional angiography significantly underestimated the distance of the stenosis from the ostium of the LAD, 29.4+/-14.5 versus 35.3+/-18.5 mm, P<0.001. Vessel curvature at the site of the lesion was overestimated by conventional angiography compared with 3-D reconstruction, 147.6+/-30.6 degrees versus 162.3+/-11.2 degrees , P<0.001, as was axial bending of the LAD owing to ventricular contraction (17.8+/-7.78 degrees vs. 8.9+/-8.9 degrees , P<0.001). No agreement was observed between two-dimensional and 3-D analysis for either curvature on lesion or axial bending assessment, with intraclass correlation coefficient values 0.155 (-0.009, 0.315) and -0.022 (-0.183, 0.174), respectively. No significant agreement was found between the two methods in the detection of on-stenosis bifurcations (1.7%, kappa=0.086, P=0.349). CONCLUSION: Conventional coronary angiography cannot provide accurate estimates of anatomical parameters, such as distance of a coronary stenosis from the ostium of the vessel, coronary artery curvature at the site of stenosis, axial deformity and bending because of ventricular contraction, and classification of bifurcations. Reconstruction of the coronary tree in 3-D space is necessary for such estimations.