The ever growing population of human beings on earth introduces the challenge of providing affordable, sustainable energy for everyone. Emerging markets, such as China, India or Brazil, quench their thirst for cheap energy by fossil fuels and nuclear power. At the same time researchers from all over the globe warn the public of the advent of a new, civilisation threatening disaster: climate change. Over the last two centuries mankind has gotten used to cheap but polluting energy provided by burning coal, gas and oil. The challenge arises in the form of the transition of our current economy towards a sustainable way of living. Renewable energy sources such as wind, tidal currents, the sun and geothermal heat have seen enourmous growth rates since the early nineties, as they are seen as the best approach to overcome this challenge. Of these renewable energy sources, wind energy is one that has received major attention. In the quest for expanding wind energy capacity, focus has shifted towards the sea in recent years. The potential energy yield is higher off-shore caused by higher average wind speeds. Maintenance and availability are key issues off-shore, due to the more complex logistics. In recent years, the price of on-shore wind energy has decreased to a level that is competitive with prices for energy from some types of fossil fuel. However, the prices for off-shore wind energy remain above the ones of fossil fuels. It is, thus, not surprising that the reduction of off-shore wind energy costs is one of the main innovation drivers within the wind industry. With the advent of off-shore wind energy more and more companies started investigating a new turbine topology called direct-drive wind turbines. This turbine type eliminates the gearbox found in other types of wind turbines, as this might lead to increased availability and lower maintenance costs. In the search for the best design of direct-drive wind turbines, every part of the turbine is investigated, analysed, measured and optimised to improve the functionality of that part. At the heart of the turbine, where the mechanical is transformed into electrical energy, is the generator. Also this component needs to be optimised with respect to weight and efficiency. This thesis aims to find the structural design that optimally utilises the mass of the generator structure to minimise deformation. This is done for the dynamic loads encountered in the generator. Special focus is given to the interaction between the structural dynamics and the magnetic field. This is important as the interaction between these two physical domains can lead to unexpected dynamic behaviour of the system. In Part I of this thesis, the modelling techniques that accurately include the interaction between the structural part of the turbine and the magnetic field in the generator are introduced. These techniques can, for the first time, predict the modal parameter changes, including damping changes, due to the interaction by forming a monolithic eigenvalue problem of the coupled system. The model neglects certain nonlinear influences on the dynamics, such as hysteresis and saturation. Its ability to predict changes of the modal parameters is validated by vibration measurements of a magneto-mechanical coupled system. Furthermore, this part develops new methods to handle huge magneto-mechanical coupled models that emerge when magnetic fields and structural dynamics of a direct-drive wind turbine are modelled. The bottleneck is the memory requirements of the monolithic formulation that makes it necessary to solve for all degrees of freedom simultaneously. Part II applies the techniques developed in Part I to the generator of the XD-115, a 5 MW direct-drive wind turbine and conducts the first two-way coupled analysis of such a generator type. The detailed dynamic analysis of the generator gives new insights in the dynamic behaviour of the generator. Furthermore, the eigenfrequencies, modes and possible causes for excitation are identified. An experimental validation of the XD-115 models was conducted using in-situ experimental and operation modal analyses. Various techniques are compared for the challenging task of exciting the rotor structure. In the second part of Part II, the loads identified during the dynamic analysis are used as load case for a structural optimisation. Topology and shape optimisation were used to identify the optimal mass distribution for the rotor structure that minimises the deformation in the air gap. This way, the weight of the structure could be reduced significantly without compromising the static and dynamic performance of the generator structure. During the optimisation the suitability and potential of topology optimisation for direct-drive wind turbines was evaluated. Although the introduced methodology can be applied to any electric machine, the implications for direct-drive wind turbine generators are most significant, as for these machines the ratio between produced torque and weight is especially high. Important influences on and encountered challenges for improving the design are collected to improve future turbine designs.