The existing theoretical modelling of thermoacoustic oscillations has been studied and applied to the case of tunnel fires. The assumptions of the theoretical model were reviewed. It was found that several assumptions are not applicable to tunnel fires and, hence, suggestions for improvement are given. The correlation which expresses the starting conditions for thermoacoustic oscillations is analysed and dependencies on different parameters are presented. The pulsations documented during the large-scale tests in the Runehamar tunnel in 2003 have also been further analysed. The measurements were compared to the theoretical limiting curve for oscillations, showing good agreement. To further study thermoacoustic oscillations in tunnels, more detailed tests in a model-scale tunnel (1:100) were performed. These tests focused on the circumstances which are required to create thermoacoustic oscillations, i. e. the fire was located at different positions along the tunnel and the air flow rate was varied. The tunnel had a length of 4 m, was 8 cm wide and 6 cm high. The results were in good accordance with the theoretical modelling but showed deviations which were most obvious at very low and high air velocities. Both the starting conditions for thermoacoustic oscillations and cases where the thermoacoustic oscillations suddenly stopped, were observed. These findings led to identification of several points in the theoretical model which need to be improved, and to the development of a strategy to avoid such pulsation in fires.