Understanding the structural response of bone during locomotion may help understand the etiology of stress fracture. This can be done in a subject-specific manner using finite element (FE) modeling, but care is needed to ensure that modeling assumptions reflect the in vivo environment. Here, we explored the influence of loading and boundary conditions (BC), and compared predictions to previous in vivo measurements. Data were collected from a female participant who walked/ran on an instrumented treadmill while motion data were captured. Inverse dynamics of the leg (foot, shank, and thigh segments) was combined with a musculoskeletal (MSK) model to estimate muscle and joint contact forces. These forces were applied to an FE model of the tibia, generated from computed tomography (CT). Eight conditions varying loading/BCs were investigated. We found that modeling the fibula was necessary to predict realistic tibia bending. Applying joint moments from the MSK model to the FE model was also needed to predict torsional deformation. During walking, the most complex model predicted deformation of 0.5 deg posterior, 0.8 deg medial, and 1.4 deg internal rotation, comparable to in vivo measurements of 0.5–1 deg, 0.15–0.7 deg, and 0.75–2.2 deg, respectively. During running, predicted deformations of 0.3 deg posterior, 0.3 deg medial, and 0.5 deg internal rotation somewhat underestimated in vivo measures of 0.85–1.9 deg, 0.3–0.9 deg, 0.65–1.72 deg, respectively. Overall, these models may be sufficiently realistic to be used in future investigations of tibial stress fracture.