The dynamic behaviour of slab and ballast tracks was investigated using measurements and calculations. Hammer impacts and train passages were analysed and measurements were made using geophones (velocity transducers) which had been time-integrated to displacements. The calculations were carried out in the frequency-wavenumber domain for multi-beam- on-continuous soil models. The characteristics of the different tracks and track elements were established in theory and by ex-periment. The frequency-dependent compliances (displacement transfer functions) showed clear rail-on-railpad resonances or highly damped track-soil resonances. Compared to the rail and sleeper, the track slab had much lower amplitudes. The slab track usually had the highest rail amplitudes due to soft railpads. Train passage yielded track displacements which were a superposition of the axle loads from the two neighbouring axles of a bogie and from the two bogies of two neighbouring carriages. This global behaviour was characteristic of the track slab of the slab track, whereas the rails of the slab and the ballast tracks behaved more locally with only one bogie of influence. The measurements agreed very well with the theory of continuous soil in the case of the six measured slab tracks and acceptably well for the six measured ballast tracks. The measurements allowed us to find appropriate model parameters and to check the models. For example, the Winkler model of the soil was found to be less appropriate because it reacted more locally.