While saccular abdominal aortic aneurysms (AAAs) are thought to be more prone to rupture than fusiform aneurysms, attempts to validate this observation have been limited by the inability to quantitatively define the three-dimensional shape of an aorta. A quantitative three-dimensional shape model may distinguish among shape classes and ultimately be useful in identifying aneurysms at risk for rupture. Three-dimensional luminal surface data of AAAs were generated from computed tomographic (CT) images of 15 patients with small aneurysms (< or =5.5 cm maximal transverse diameter). The centerline was used to construct a shape classification based upon the orthographic projection of the centerline about its central axis. The extent and direction of the individual deviations were quantified as areas on the plane of projection to create a shape classification. Hierarchical cluster analysis was used to verify distinct shape classes. A tortuosityindex was calculated as a function of the centerline projection. AAA shape was calculated as a tortuosityindex and classified into distinct classes of minimal or increased three-dimensional tortuosity. Thrombus could change the tortuosityindex or shape classification of an aneurysm. In several patients with serial CT scans, the tortuosityindex changed over time and was correlated with rupture; in three AAAs that ruptured the mean tortuosity increased 29% whereas the mean transverse diameter increased 3.3%. Expanding AAAs develop specific, quantifiable shapes that can be expressed as a quantitativetortuosityindex that may be relevant to their natural history. The three-dimensional features of this shape model provide a novel and potentially clinically relevant adjunct to maximal transverse diameter. Larger studies are needed to correlate the tortuosityindex with finite element models and the ability to predictaneurysmrupture.