In this paper the performance of several modal parameter estimation techniques is compared by considering a simple mechanical structure as test bed. Specifically, among Input/Output (I/O) techniques, a frequency domain poly-reference method is considered. Among Operational Modal Analysis (OMA), Frequency Domain Decomposition and poly-reference least square complex frequency-domain techniques are used. I/O technique and OMA are performed using the measurements due to a random shaker excitation. In the first case the responses and the input force are used to calculate the FRFs. On the contrary, OMA is performed by using only the set of measured responses, by supposing unknown the input force. Note that the OMA hypothesis of a random distributed force is not observed. In order to characterise the performance of the examined methods, the results obtained by the OMA are compared to the result of the I/O technique.