This work present an application of the Kalman filter with a recursive estimator to determine the excitation loads of linear dynamic systems. The loads were estimated from the measured dynamic responses through an inverse method. The estimator employed a least squares algorithm to compute the magnitudes and, therefore, the onset time histories of the loads. The practicability and accuracy of the proposed method were verified with laboratory tests from with the input loads of a cantilever plate were estimated from the measured dynamic responses. The dynamic characteristics of the cantilever plate the modal testing, with frequency response function identifying the system model parameters, and setting its mathematic model. In the experimental study, the plate was subjected to periodic sinusoidal and rectangular loads. The estimation results of the acting loads are qualitatively and quantitatively good in all cases tested.