Linear dynamic finite element analysis can be considered very reliable today for the design of aircraft engine components. Unfortunately, when theses individual components are built into assemblies, the level of confidence in the results is reduced, since the joints in the real structure introduce nonlinearity that cannot be reproduced with a linear model. Certain types of nonlinear joints in an aircraft engine, such as underplatform dampers and blade roots, have been investigated in great detail in the past, and their design and impact on the dynamic response of the engine is now well understood. With this increased confidence in the nonlinear analysis, the focus of research now moves towards other joint types of the engine which must be included in an analysis to allow an accurate prediction of the engine behaviour. One such joint is the bolted flange, which is present in many forms on an aircraft engine. Its main use is the connection of different casing components to provide the structural support and gas tightness to the engine. This flange type is known to have a strong influence on the dynamics of the engine carcase. A detailed understanding of the nonlinear mechanisms at the contact is required to generate reliable models and this has been achieved through a combination of an existing non-linear analysis capability and an experimental technique to accurately measure the nonlinear damping behaviour of the flange. Initial results showed that the model could reproduce the correct characteristics of flange behaviour, but the quantitative comparison was poor. From further experimental and analytical investigations it was identified that the quality of the flange model is critically dependent on two aspects: the steady stress/load distribution across the joint and the number and distribution of non-linear elements. An improved modelling approach was developed which led to a good correlation with the experimental results and a good understanding of the underlying nonlinear mechanisms at the flange interface.