Using paired humeri with a midshaft osteotomy, the biomechanical stiffness of four intramedullary internal fixation devices were compared with each other, a dynamic compression plate, and with the intact bone. In posterior and lateral bending and in torsion, flexible intramedullary pin fixed humeri (Enders and Hackethal) performed similarly and were less stiff than intact specimens were. Interlocking intermedullary nail constructs (Russell-Taylor and Seidel) also tested similarly to each other, and were stiffer than the flexible pins in all bending tests. Compared with the intact humerus, interlocking nails were stiffer in torsion, but in bending they more closely simulated the stiffness of the bone.