Twenty-four matched pairs of fresh-frozen humeri from human cadavera were divided randomly into four groups, in order to determine the most biomechanically desirable construct for the prophylactic fixation of impending fractures of the distal third of the humerus. Group I comprised intact humeri and matched humeri in which a 50 per cent lateral, semicylindrical cortical defect of the distal third had been created, resulting in a reproducible model of an impending fracture due to a lytic defect involving 50 per cent cortical disruption at the distal end of the humeral medullary canal. In Group II, such a lateral defect was created in both the right and the left, matched humeri. Group III was composed of humeri in which the defect had been fixed prophylactically with a single plate and the contralateral humeri, which had been treated with double-plating. Group IV comprised specimens in which the defect had been fixed with double-plating as well as those fixed with Rush rods. The fixation of each specimen in Groups III and IV was supplemented with bone cement. Each specimen was tested in torsion to failure, and the resulting peak torque, torsional stiffness, and total energy absorbed were analyzed for each group. The Group-I specimens that had a defect had a significantly lower (p < 0.05) peak torque, torsional stiffness, and total energy absorbed than the intact specimens; all of the specimens with a defect failed at the defect, and all of the intact specimens failed proximally. In Group II, there was a high side-to-side association with respect to peak torque, torsional stiffness, and total energy absorbed.(ABSTRACT TRUNCATED AT 250 WORDS)