Abstract: The dip-angle-domain common-image gather (DDCIG) is a key tool to separate the diffraction and reflection imaging results. Reflectors with different spatial geometries produce different responses in DDCIGs. Compared with Kirchhoff migration, Gaussian beam migration (GBM) is more effective and robust to overcome the multipathing problem. As a ray-based method, it has explicit angle information naturally during the propagation. We have developed a 3D DDCIG computational method using GBM, which obtain both the imaging result and angle-domain gathers with only one pass of calculation. The angle-gather computation is based on geometrical optics, and multiple angle conversions are implemented under the rules of space geometry, which helps to avoid rounding errors and improve accuracy. Additionally, the multi-azimuth joint presentation strategy is proposed to describe the characteristic of omnidirectional dip angles using a finite number of gathers. After using a 2D model to illustrate application advantages of DDCIG, we apply the proposed method to two 3D models to test its feasibility and accuracy. A field data example further demonstrates the adaptability of our method to seismic imaging for a land survey.