This paper addresses the problem of estimating left ventricle rigid and nonrigid motions from three-dimensional (3D) dynamic CT data via a model-based approach. We have developed a two-stage scheme to estimate the motion and deformation of the left ventricle. In the first stage, the rigid motion of the left ventricle is computed from the position and orientation change of an object-centered coordinate system. In the second stage, the global deformations of the left ventricle are estimated by fitting the digital surface to su-perquadric modeling primitives which we use to model the time-varying and deformable left ventricle shape. Deformation operations, such as tapering and bending, are applied to the basic superquadric modeling primitives in order to capture the complex nature of the left ventricle shape and nonrigid motions. The preliminary results obtained seem promising, since they match well to the apparent motion pattern of the left ventricle.