The estimation of vertebral fracture risk in individuals with suspected osteopenia is commonly based on measurements of lumbar spine bone density. The efficacy of vertebral size and deformity, as assessed by vertebral morphometry, in the prediction of fractures has been less studied. In an ex vivo investigation the regional relationships between vertebral size, vertebral deformity, bone density and compressive strength throughout the thoracolumbar spine were examined. In 16 vertebral columns (T1-L5) the bone mineral content (BMC) and bone mineral density (BMD) of each segment were measured using lateral projection dual-energy X-ray absorptiometry, and the vertebral cancellous density (VCD) and mid-vertebral cross-sectional area (CSA) measured using quantitative computed tomography. Vertebral body heights were determined from mid-sagittal CT scans, and vertical height ratios calculated for each segment. The failure load and failure stress of the isolated vertebral bodies were determined using a material testing device. Separate analyses were performed for the upper (T1-4), middle (T5-8) and lower (T9-12) thoracic, and lumbar (L1-5) segments. In all regions, failure load was strongly correlated with BMD (r = 0.82-0.86), moderately correlated with VCD (r = 0.60-0.71) and vertebral height (r = 0.22-0.49), and poorly correlated with the height ratios (r = 0.04-0.33). Failure stress was best predicted by BMD (r = 0.73-0.78) and VCD (r = 0.70-0.78) but was poorly correlated with all morphometric variables (r = 0.01-0.33). The segmental correlations between BMD and VCD ranged form r = 0.49 to r = 0.79. For all regions, BMD and VCD were included in the stepwise regression models for predicting failure load and failure stress. Either the mid-vertebral height or CSA were included in all the failure load models, while mid-vertebral height was included in only one of the failure stress models. The results suggest that vertebral deformity and size (as assessed by vertebral morphometry) make only a minor contribution to the prediction of vertebral strength additional to that provided by bone densitometry alone. The consistent regional relationships between variables appear to support the practice of global fracture risk assessment based on lumbar spine densitometry.