There are differences in the dynamic behaviour of a machine tool between standstill and process because of preloads, process damping and gyroscopic moments. The standard method to characterize the dynamic behaviour, the experimental modal analysis (EMA), is performed only in the standstill and neglects these effects. In contrast, operational modal analysis (OMA) makes possible a characterisation of the dynamic behaviour under working conditions. Nevertheless, the performance of OMA is based on some assumptions. In this paper several key aspects considering these requirements regarding to machine tools are discussed. These presumptions affect the time-invariance during the measurement, the suitability of a milling process to excite the machine adequately and the influences of the different identification methods to the results. The obtained results of these investigations will be used in further studies to make OMA applicable for machine tools.