Abstract: Lattice structures manufactured by metallically additive manufacturing have incrementally been applied in aerospace, national defense, and military industries, but the corresponding defect detection methods are still lacking. The industrial CT detection method used currently is costly and time-consuming, and incapable of meeting the detection requirements for a large number of lattice structures. In this paper, based on ultrasonic resonant spectroscopy, the Mahalanobis-Taguchi system classification algorithm is adopted to study the non-destructive testing of missing struts in lattice-structure parts. The Mahalanobis distance correlates positively with the number of missing struts, which suggests that the detection requirements for abnormal parts can be flexibly adjusted by moving the classifying threshold of Mahalanobis distance to change the strictness of the detection index. Simulation and experimental results show that this method can accurately identify strut-missing parts according to the magnitude of the Mahalanobis distance. This study provides a reference value for ensuring the integrity and reliability of lattice-structure parts produced by additive manufacturing.