The complexity of a room model affects to the computational resources required to model its acoustics by means of geometrical acoustics modeling methods. Thus, a method for reducing the geometry of the room models is presented for static room geometries. The topology of the model is simplified in a process where the model is first decomposed into a volumetric structure. The surface is reconstructed by utilizing this structure, and subsequently simplified by merging coplanar regions. The results of the method are verified by extracting room acoustical attributes from the original and reduced models with the ODEON room acoustics prediction software. It is shown that the most important acoustic properties have been preserved, even with relatively high reduction rates.