Sustainable railway systems with less disturbance are highly desired. Insulated rail joint (IRJ) is one important structural component in a railway track. IRJs serve the railway system with two crucial functions: dividing track into sections for signal control and enable the detection of broken rails. An IRJ generally consists of two fishplates, insulation layers between the rails and the fishplates, and an end-post layer between two rails. The fishplates are assembled using pre-tensioned bolts. Although the IRJ plays an important role in the railway system, it introduces discontinuities in stiffness and geometry to wheel-rail rolling contact and is thus considered one of the weakest parts of the railway track. Maintenance of IRJs has been an issue of great concern for the rail operators worldwide. Over the past few years many studies have been conducted on monitoring, modelling and analysis of IRJs. Both numerical and experimental studies have been performed in order to gain better understanding of the behaviour of IRJs under static and dynamic loads. Considering that dynamic wheel-rail impacts occur at IRJs, the dynamic behaviour of IRJs should be carefully examined. This M.Sc. thesis studies the dynamic behaviour of the IRJ experimentally and numerically. Two three-dimensional finite element (FE) track models with different types of IRJ, i.e. a new NRG-joint and normal joint, are developed. Rails, fishplates, insulation layer and sleepers are represented in the models as detailed as possible, while the rail-pads and the ballast are simplified as spring and damper elements. The NRG-joint consists of two fishplates and six pretensioned bolts, while the normal joint consists of two shorter fishplates and four bolts. Implicit FE analyses are first performed to calculate the initial stresses and displacements of the track models, which are then used as initial conditions for the explicit dynamic FE analysis. Field hammer tests are performed on the track line between Zwolle and Meppel. The tests are used on one hand to calibrate the stiffness and damping parameters involved in the rail-pad and ballast models, and on the other hand to validate the dynamic behaviour reproduced by the FE IRJ models. The comparison between the simulation and the measurements shows reasonable agreement of the results. Based on the validated FE models, optimisation strategies of the dynamic behaviour of the IRJ are proposed by a parametric study. The varied parameters include stiffness and damping of the rail-pads and the ballast as well as sleeper span. Implementation of non-uniform sleeper span seems effective for reducing pinned-pinned resonance. Finally, conclusions are drawn and recommendations for further research are made.