Abstract-Shape representation methods play an important role in 3D shape recognition system. Three-dimensional shape recognition is widely used in 3D search engines, gravitational field, medical imaging, computer vision and face recognition. In this paper we propose an ellipsoidal shape representation technique for 3D shape recognition. We present some experimental and comparison results of our approach for shape matching using a standard database, Princeton Shape Benchmark. The effectiveness of our proposed algorithm is measured using nearest neighborhood. We then introduced a new idea which is a possible extension of the proposed approach and evaluate the results against human observation.