Shimmy, or wobble, denotes the unstable self-excited oscillatory motion of the wheel around the steering axis and can be found in vehicles with castor wheels. For bicycles this oscillatory motion is normally in the range between 5 - 13 Hz. Shimmy could be frightening, dangerous and even fatal if not controlled in time. The introduction of electric bicycles led to a growing occurrence of shimmy at low speeds and thereby raised the demand for a better understanding of this phenomenon. Desirably, it is known which parameters must be adjusted to prevent shimmy in future bicycle designs. However, the literature study prior to this thesis revealed that this is rather complex and many contradictory results were found. This raised the demand for a model that could predict the behaviour, and hence the influence of parameters on the wobble mode. The most widely used bicycle model nowadays, the Whipple-Carvallo bicycle model can not predict shimmy behaviour and it became clear that the bicycle model must be extended with additional degrees of freedom to account for frame compliances. The position and orientation of this degree(s) of freedom affect the predictions. However, it was unknown which additions should be made for correctly predicting shimmy and the predictions could not be validated with measurement data. Within this study, the shimmy vibrations of four female and two male bicycle models are measured for three different drivers with 16 to 19 accelerometers, a steer angle sensor and a speedometer. The frequency and amplitude growth are extracted as a function of forward speed. The corresponding bicycle-driver parameters are also measured. Both results accommodate the validation of bicycle models that can predict shimmy behaviour. The typical mode of vibration is also extracted and it was observed that front fork, rear frame and rear rack bending is apparent. The driver also oscillates in the shimmy frequency and relative motions between the driver and bicycle were observed. The deformations of the rear frame appeared to be complex and can be approximately described by deflections around two lines in the frame; One more vertical and one more horizontal line. The mode shapes give directions for improving the model-based shimmy predictions.