One hundred and twenty patients had an allograft reconstruction of the femur, tibia, or humerus. Of a total of 183 allograft-host junctions, eight-three were fixed with a plate; ninety-eight, with an intramedullary rod; and two, with screws alone. There was no significant difference between the rate of union after fixation with a plate and that after intramedullary fixation (p = 1.00). However, fixation with a plate was associated with a higher rate of fracture of the allograft (p < 0.0001). Some problem related to the internal fixation of the allograft was identified at eighteen of the twenty junctions that did not heal. There was a significant association (p < 0.001) between a problem in the achievement of stable fixation and the development of a non-union at the allograft-host junction.