Purpose: To examine whether x-ray vector radiographic (XVR) parameters could predict the biomechanically determined vertebral failure load. Materials and Methods: Local institutional review boards approved the study and donors provided written informed consent before death. Twelve thoracic vertebral bodies were removed from three human cadavers and embedded in resin. XVR measurements were performed by using a Talbot-Lau grating interferometer with the beam direction in anterior-posterior and lateral direction. The mean anisotropy and the mean local average scattering power were calculated for a region of interest within each vertebra. Trabecular bone mineral density (BMD) was determined in each vertebra by using a clinical multidetector computed tomographic scanner. Failure load of the vertebral bodies was determined from destructive biomechanical tests. Statistical analyses were performed with statistical software with a two-sided Pvalue of .05 to calculate Pearson correlation coefficients and multiple regression model. Results: Statistically significant correlations (P < .05) for failure load with XVR parameters in the lateral direction (r = 20.84 and 0.68 for anisotropy and local average scattering power, respectively) and for failure load and anisotropy in anteroposterior direction (r = 20.65) were found. A multiple regression model showed that the combination of the local average scattering power in lateral direction and BMD predicted failure load significantly better than BMD alone (adjusted R = 0.88 compared with 0.78, respectively; P < .001). Conclusion: The study results imply that XVR can improve the prediction of osteoporosis.