Fractures of the vertebral body of the human spine represent a serious health care problem to the aging populations of the United States and other nations. At least 500,000 vertebral fractures occur in the U.S. each year, and an estimated one in four women over age 65 has had one or more fractures. Vertebral fractures are associated with back pain, loss of height, spinal deformity and increased mortality. The objective for this thesis was to address several questions related to vertebral fracture etiology by investigating the behavior both of vertebral trabecular bone (the primary constituent of the vertebral body) at the microstructural level and of the whole vertebral body at the continuum level. In particular, 1) What effect do microstructural defects have on the mechanical properties of trabecular bone? and 2) Can the failure behavior of vertebral bodies at the whole bone level be predicted accurately using finite element analysis? Using a novel approach for modeling the microstructure of vertebral trabecular bone, two-dimensional finite element models were developed that predict realistic failure behaviors. Analyses performed with these models indicate that the presence of even a small number of defects in trabecular bone strongly degrade its mechanical properties, and suggest that maintaining the number of trabeculae (i.e., preventing defects) is essential for effective treatment of osteoporosis. At the continuum level, the accuracy of computed tomography (CT)-based finite element analysis for predicting the failure behavior of bone structures was investigated by comparisons between experimental and predicted behaviors of heterogeneous sections of vertebral bone. Results indicate that CT-based finite element models can be used to predict accurately both failure load and fracture patterns in these simple bone structures. These methods were extended to the whole bone level with similar success. The failure loads of three vertebral motion segments were predicted within 10% of the measured failure loads, and the predicted fracture patterns matched those observed experimentally. The methods developed in this study represent important new tools for research in the mechanics of age-related vertebral fractures.