Some methods in experimental modal analysis rely on a finite set of modes and they neglect the higher modes. However, this approach causes a truncation of the modal decomposition and the modal truncation introduces errors of unknown magnitude. In this paper the effect of modal truncation is investigated on a test specimen in the laboratory. It is found that the system response is dependent of the frequency and the distribution of the load. Modal truncation can introduce significant errors if the set of mode shapes does not efficiently span the spatial distribution of the load.