In order to assist surgeons during surgery on moving organs, e.g. minimally invasive beating heart bypass surgery, a master-slave system which synchronizes surgical instruments with the organ's motion is desired. This synchronization requires reliable estimation of the organ's motion. In this paper, we present a new approach to motion estimation based on a state motion model for a partition of the heart's surface. Its motion behavior is described by a partial differential equation whose input function is assumed to be periodic. An estimator is used on one hand to predict future model states based on reconstruction of the input function and on the other hand to incorporate noisy spatially discrete measurements in order to improve state estimation. The model-based motion estimation is evaluated using a simple heart simulator. Measurements are obtained by reconstructing 3D position of markers on a pulsating membrane by means of a stereo camera system.