We represent a shape representation technique using the eigenfunctions of Laplace-Beltrami (LB) operator and compare the performance with the conventional spherical harmonic (SPHARM) representation. Cortical manifolds are represented as a linear combination of the LB-eigenfunctions, which form orthonormal basis. Since the LB-eigenfunctions reflect the intrinsic geometry of the manifolds, the new representation is supposed to more compactly represent the manifolds and outperform SPHARM representation. We demonstrate the superior reconstruction capability of the representation using cortical and amygdala surfaces as examples.