Subaerial landslide-tsunamis (impulse waves) are generated by mass movements such as landslides, rock falls or glacier calving interacting with a water body. Preliminary landslide-tsunami hazard assessment is commonly based on empirical equations derived from wave channel (2D) or wave basin (3D) experiments. It is crucial to select the most appropriate set of empirical equations for a particular case as the difference in the far-field wave height between 2D and 3D may exceed an order of magnitude. The present study systematically investigates the effect of the water body geometry on the wave characteristics. Physical model tests were conducted in 2D and repeated in 3D, involving two water depths, three rigid slides and different subaerial slide release positions. The waves were found to decay in 2D considerably slower with distance x ‒0.30 than in 3D with radial distance r ‒1.0. The 3D wave heights in the slide impact zone can be identical large as in 2D for a large slide Froude number F, relative slide thickness S and relative mass M. However, for small F, S and M, the 3D waves are considerably smaller, both in the near- and far-field. Empirical equations are presented to transform wave parameters from 2D to 3D. One 2D-3D test pair, involving a solitary-like wave, is investigated in detail regarding the slide kinematics, water surface elevations and slide-water interaction power. This power is derived from pressure measurements on the slide front and the slide kinematics. The identical test pair is then used to calibrate the Smoothed Particle Hydrodynamics SPH code DualSPHysics and to numerically investigate the wave features in five intermediate geometries between 2D and 3D. For a “channel” geometry with diverging side wall angle of 7.5°, the wave amplitudes along the slide axis were found to lie approximately halfway between the values observed in 2D and 3D. At 45°, the values are practically identical to those in 3D. These findings support preliminary landslide-tsunami hazard assessment.