We reviewed 64 patients in whom 66 acetabula had been reconstructed with either the Muller ring (46) or the Burch-Schneider anti-protrusio cage (20) at a mean follow-up of five years. Five hips had been revised a second time for loosening, all after a Muller ring had been used for a medial segmental defect (2), ungrafted cavitary defects (2) or after resorption of a block graft (1). The use of bone grafts with the implants reduced the incidence of failure from 13% to 6% and of circumferential radiolucent lines at the bone-implant interface from 39% to 2%. The Muller ring is indicated for acetabula with isolated peripheral segmental defects or cavitary defects confined to one or two sectors. The Burch-Schneider cage should be used for medial segmental defects, extensive cavitary defects and combined deficiencies. Defects should be reconstituted with bone graft rather than cement.

We treated 24 patients with nonunion of tibial shaft fractures by locked intramedullary nailing, 18 by open and six by closed techniques. Union was achieved in 22 patients, failing only in two patients with active infection. Locked nailing prevented recurrence of deformity and allowed the patients to mobilise without external support. Supplementary bone grafting was essential only for major defects.

The incidence of refracture following the removal of screws and plates from the diaphyses of 115 forearm bones in 80 patients has been studied. Refracture occurred in four adult patients as a result of minimal trauma, in two patients at the original fracture site after premature plate removal, at the site of a countersunk interfragmentary screw in one and at the original fracture site in another who had required three operative procedures to achieve 'union'. It is suggested that refracture could have been avoided in at least two of these patients. If the 3.5 mm plating system has been used, the incidence of refracture should be minimal.

Bone weakness leading to refracture is a recognised complication of the removal of rigid fixation plates. We have used partially demineralised rabbit tibiae to simulate atrophic changes and to determine whether weakness is due to atrophy or to residual screw holes. Partial demineralisation and a screw hole each reduced maximum bending moment. However, energy absorbing capacity was little affected by demineralisation, but was reduced to 50% by a single drill hole. Residual screw holes are a considerably more important cause of bone weakness after plate removal than is cortical atrophy.

We used single-photon absorptiometry to assess the forearm bones after the removal of internal fixation plates in 14 patients. We found convincing evidence of cortical atrophy in only one patient, in whom the plates had been removed prematurely after only 16 months. It is suggested that such plates should be retained for at least 21 months, to allow bone density to return to its prefracture level. The recommendations of the AO/ASIF group are supported.