The authors estimated noninvasively the wall stress distribution for actual abdominal aortic aneurysms (AAAs) in vivo on a patient-to-patient basis and correlated the peak wall stress (PWS) with various geometrical parameters. They studied 39 patients (37 men, mean age 73.7 +/- 8.2 years) with an intact AAA (mean diameter 6.3 +/- 1.7 cm) undergoing preoperative evaluation with spiral computed tomography (CT). Real 3-dimensional AAA geometry was obtained from image processing. Wall stress was determined by using a finite-element analysis. The aorta was considered isotropic with linear material properties and was loaded with a static pressure of 120.0 mm Hg. Various geometrical parameters were used to characterize the AAAs. PWS and each of the geometrical characteristics were correlated by use of Pearson's rank correlation coefficients. PWS varied from 10.2 to 65.8 N/cm2 (mean value 37.1 +/- 9.9 N/cm2). Among the geometrical parameters, the PWS was well correlated with the mean centerline curvature, the maximum centerline curvature, and the maximum centerline torsion of the AAAs. The correlation of PWS with maximum diameter was nonsignificant. Multiple regression analysis revealed that the mean centerline curvature of the AAA was the only significant predictor of PWS and subsequent rupture risk. This noninvasive computational approach showed that geometrical parameters other than the maximum diameter are better indicators of AAA rupture.