Abstract: This paper firstly investigates 3D nose, one of the most prominent parts of the human face, as a novel biometric modality for human identity recognition.An effective 3D nose shape recognition approach is proposed, which consists of three key components:automatic extraction of the nose region based on feature points, alignment of nose shapes by ICP, and similarity measure for noses.The experiments are conducted on FRGC 1.0 database, achieving 3.15% of equal error rate and 92.1% of rank-one recognition rate.Experimental results show that 3D nose shape has comparably capability of discrimination.