Abstract: Monitoring (geotechnical) constructions is often based upon displacement measurements. However, these measurements do not offer information about the stiffness behaviour of a soil-structure system. A loss of stiffness might be observed as a decrease of the system’s eigenfrequencies. This research investigates if monitoring of ambient vibrations can be used to observe a change in the system’s stiffness. Stiffness monitoring of structural parts (e.g. steel and concrete beams) using vibrations is already common. These implementations are based on measuring natural frequencies and mode shapes. Any change in structural stiffness results in a change in these modal characteristics of the structure. A technique similar to this, but operating in the lower frequency range (i.e. below 300 hertz), is already used to derive the shear elasticity of soil. These techniques are known as seismic methods, and they record body and surface waves. The denser and stiffer the layer of the strata is, the faster it vibrates and the faster the phase velocity of the recorded waves will be. This provides an estimate of the strength of the soil and its ability to resist permanent deformation (i.e. its elastic behaviour). It is also used to find boundaries between different soil layers. In this research, the possibility of monitoring a relative change in stiffness during construction works is investigated. By a relative change is meant the change in stiffness with respect to the initial stiffness, expressed as a percentage. The initial stiffness will be coupled to the initial eigenfrequency of the system. A changed eigenfrequency can then be coupled to a percentage of this initial stiffness. The soil-structure system used for the analytical and empirical part of this research is part of a railway bridge in Nijmegen. In Nijmegen, diaphragm walls are constructed to a depth of more than 20 meters, surrounding the old pillars of this bridge. It is assumed that, during construction works, there will be a change in the system’s stiffness due to the installation of the diaphragm walls. The eigenfrequencies of the soil-structure system are determined by continuous vibration monitoring of ambient vibrations, where the ambient vibrations are caused by the railway traffic. When the stiffness k decreases, the eigenfrequency of the system should also decrease. Two models have been analysed to simulate the bridge: a single and a double mass-spring model. Multiple parameters of these mass-spring models are modified in order to determine which parameter influences the eigenfrequency of both the soil and the structural part of the system. From the mass-spring model it follows that the dominant frequency in the lower frequency range, between 5 and 15 hertz, represents the eigenfrequency of the soil. The dominant frequency in the higher frequency range, between 40 and 50 hertz, represents the eigenfrequency of the structure. With a changing stiffness of the construction, the eigenfrequency between 40 and 50 hertz changes significantly while the change in eigenfrequency around 10 hertz is insignificant. When the stiffness of the soil decreases, the eigenfrequency around 10 hertz decreases significantly while the eigenfrequency between 40 and 50 hertz remains almost unchanged. With the mass-spring model it is also concluded that only a change in stiffness relative to the initial stiffness can be monitored. A Fast Fourier Transform is used to convert the measured data into a frequency spectrum. When there is a phase difference between the first and the last data point a so called leakage occurs. Since it is impossible to determine the phase of the signal when dealing with ambient vibrations, a phase difference cannot be avoided. Due to leakage, the velocities in the frequency spectrum do not correspond well to the real velocities. Actual velocities may be more than 30% higher than the velocities obtained after a Fast Fourier Transform. With the recorded datasets of both the author, in cooperation with the Municipality of Rotterdam, and Fugro GeoServices B.V. it is investigated if the eigenfrequencies are changing during the construction works. The initial data is compared with the data recorded during and after the construction works. The dataset recorded by the author contains continuous vibration measurement in three directions, recorded by 10 geophones with a sampling frequency of 1000 hertz. The datasets are recorded on two different days. The first day represents the initial phase. The second day represents the construction phase. From these measurements it can be concluded that different types of trains do not have an influence on the observed eigenfrequencies. On the other hand, they do have an influence on the magnitude of the recorded vibration. In between the two days of measurement, hydraulic jacks were installed in between the girders and the pillar of the bridge to correct the settlements that occurred during the construction of the diaphragm walls. These jacks have made the joint in between the girder and the pillar more rigid. Due to this, the girder is acting stiffer than before. From the dataset it can be concluded that the eigenfrequency of the structure increases significantly after installation of the jacks. The frequency peak values representing the pillar and the girder increase. The eigenfrequency of the soil remains almost unchanged after installation of the jacks. This conclusion is consistent with the results that followed from the analytical model, as described above. With the dataset recorded by Fugro GeoServices B.V. it is possible to analyse and compare the results of the initial phase, the construction phase and the post phase. For the analysis only the recorded traces are used. These traces contain continuous vibration measurements during 2 seconds, with a sampling frequency of 1024 hertz. The lower frequency range, between 0 and 25 hertz, of multiple monitored traces is analysed and compared. From these results, it can be concluded that a change in stiffness of the soil can be observed by a shift in eigenfrequencies. A decrease in eigenfrequency compared to the initial measured eigenfrequency is observed during the construction works. The decrease is small, but comparable to the decrease that was expected beforehand. In the post phase, when the construction works are finished, the eigenfrequency increases again. This leads to the conclusion that the stiffness has recovered again. Observing a change in stiffness of a soil-structure system by shifts in eigenfrequencies is possible, but only a relative stiffness change can be observed (i.e. the change in stiffness with respect to the initial stiffness). It is possible to monitor the shifts in eigenfrequencies by measuring ambient vibrations. It should be noted that with this monitoring system it is not possible to monitor settlements. The outcome of this research is relevant for stiffness monitoring of constructions. For projects where the deformation of a construction is rather irrelevant if the stiffness is not being influenced significantly, a vibration monitoring system which monitors shifts in eigenfrequency can offer information about the dynamic response (i.e. stiffness behaviour) of the construction.