Seismic diffractions, which are produced by heterogeneities smaller than seismic wavelength, have proved useful in a number of 2D and 3D applications. Among the challenges hampering the use of diffractions in seismic data processing is the sensitivity of diffraction separation after prestack migration to velocity errors. Therefore, it is essential to properly account for velocity anisotropy in diffraction separation and imaging.Here, we analyze the signature of diffractions in orthorhombic models, which are typical for many subsurface formations. Tip and edge diffraction events are modeled for a tetrahedron embedded in an orthorhombic medium. We construct 3D dip-angle common-image gathers (CIGs) as a function of azimuth and migration dip using the actual orthorhombic velocity field as well as a purely isotropic velocity model. The modeling results provide insight into the influence of velocity errors on the diffraction moveout in different azimuths in the dip-angle domain. This work should aid in constructing diffraction-based depth images using 3D anisotropic velocity models.