Adhesive bonding offers a simple and efficient way of joining structural components without weakening them by holes or welding. The objective in the thesis is to examine the mechanical response of adhesively bonded joints, and test different failure prediction approaches with analytical formulas and finite element models. The thesis starts with an introduction of the most relevant theory for adhesively bonded joints. Then, a newly developed model to predict the fracture load of bonded overlap joints using a fracture mechanics approach is tested against current strength-of-materials capacity models. This testing is mainly performed with the use of finite element models, but the development of an FE model with cohesive elements, that can be used to test the fracture mechanics approach, was a challenging process. So the development of this model is described in detail. A comparison of experimental results with predictions by current strength-of-materials capacity models shows a discrepancy between the theoretical predictions (from the FE models) and the experimental results. In contrast, the new model based on a fracture mechanics approach shows a good agreement with experimental results. The FE model based on the fracture mechanics approach is then used to see if it can tell how the fracture propagates in bonded joints, and the result is consistent with experimental results. At the end of this thesis, a recommendation of capacity models in bonded joints is made based on the results in this thesis and relevant reports.