Abstract: Head-mounted three-dimensional (3D) display is a common presenter for virtual reality, augmented reality and metaverse. Currently, the image quality evaluation methods and systems for head-mounted 3D display are insuffient in quantity and low in accuracy. In this work, a 3D image quality evaluation theory based on the human visual system (HVS) is proposed. A 3D image quality evaluation method and the corresponding system are designed. Some parameters including depth reconstruction, distortion, ultramate resolution, and field of view are quantitatively evaluated. In this paper, the depth reconstruction is tested based on binocular principle, the distortion is expressed by curvatures and polynomial transformations of specific points, ultramate resolution is evaluated by MTF and valley-to-peak ratio of patterns, and the field of view is measured based on pinhole imaging method. In the test of two head-mounted display devices, the relative error of depth reconstruction is smaller than 3.5%, the distortion is smaller than 3%, and the field of view angles are 77°1′44″ and 86°56′26″, respectively. The evaluation results are highly consistent with the subjective perception of the human eyes. The propsed method and system have high applicability, which are expected to be widely applied in the fields of qualitive evaluation and design optimization for head-mounted 3D display.