Résumé : Background: One of most accepted principles for treating Kienböck's disease before wrist degeneration settles in is to decompress the lunate by an osteotomy. Several osteotomies have been proposed since 1935. However, they are based on biomechanical hypotheses that are sometimes conflicting: This study compares the decompression effect of radius transverse shortening, radius lateral closing and medial closing wedge osteotomies, capitate shortening – with and without hamate shortening – and a Camembert-type radius wedge osteotomy with and without ulnar head shortening according to Sennwald. Methods: We built a 3D wrist model using finite elements that included the metacarpal, carpal and forearm bones. All wrist ligaments and Triangular Fibrocartilage Complex were incorporated in the simulation. Load was applied on the metacarpals with the forearm bones fixed. We then applied the different osteotomies to the model. Findings: When load was applied to the wrist, the osteotomies that best unloaded the lunate were the capitate shortening osteotomy combined with hamate shortening and the Camembert osteotomy combined with ulna shortening; the latter was the only osteotomy that completely unloaded the lunate. Interpretation: We think the association of the radius Camembert osteotomy and ulna Sennwald's shortening osteotomy is the most effective procedure to propose in Kienböck's disease.