Patient-specific measures derived from quantitative computed tomography (QCT) scans are currently being developed as a clinical tool for vertebral strength prediction. QCT-based measurement techniques vary greatly in structural complexity and generally fall into one of three categories:​ (1) bone mineral density (BMD), (2) “mechanics of solids” (MOS) models, such as minimum axial rigidity (the product of axial stiffness and vertebral height), or (3) three-dimensional finite element (FE) models. There is no clear consensus as to the relative performance of these measures due to differences in experimental protocols, sample sizes and demographics, and outcome metrics. The goal of this study was to directly compare the performance of QCT-based assessment techniques of varying degrees of structural sophistication in predicting experimental vertebral compressive strength. Eighty-one human thoracic vertebrae (T6–T10) from 44 donors cadavers (F = 32, M = 12;​ 85 ± 8 years old, max = 97 years old, min = 54 years old) were QCT scanned and destructively tested in uniaxial compression. The QCT scans were processed to generate FE models and various BMD and MOS measures, including trabecular bone mineral density (tBMD), integral bone mineral density (iBMD), and axial rigidity. Bone mineral density was weakly to moderately predictive of compressive strength (R² = 0.16 and 0.62 for tBMD and iBMD, respectively). In vitro vertebral strength was strongly correlated with both axial rigidity (R² = 0.81) and FE strength measurements (R² = 0.80), and the predictive capabilities of these two metrics were statistically equivalent (p > 0.05 for differences between FE and axial rigidity). The results of this study indicate that non-invasive predictive measures of vertebral strength should include some level of structural sophistication, specifically, gross geometric and material property distribution information. For uniaxial compression of isolated vertebrae, which is the current biomechanical testing paradigm for new non-invasive strength assessment techniques, QCT-based FE and axial rigidity measures are equivalent predictors of experimental strength. However, before abandoning the FE method in favor of more simplistic techniques, future work should investigate the performance of the FE method versus MOS measures for more physiologically representative loading conditions, e.g., anterior bending or in situ loading with intervertebral discs intact.