The purpose of this study is to authenticate a surface from a set of surfaces based on 3D acquisitions in the context of corneal biometric application. The key idea is to quantify the difference between surfaces, and to define a threshold to determine if compared surfaces belong to the same person. This threshold depends on the normal variation between acquisitions of a same subject, and on the measurement noise. The proposed approach consists in comparing coefficients from a spherical harmonics decomposition. First we present basic concepts about corneal measurement and previous works related to corneal surface comparison. Then spherical harmonics are presented, followed by a description of our method. Finally, our approach is compared to the existing one: better false-match rate and false-non-match rate have been observed with the proposed method.