Background: Disruption of the extensor mechanism is an uncommon but devastating complication of total knee arthroplasty. Several techniques for reconstruction of the extensor mechanism after total knee arthroplasty have been reported, but we do not know of any study in which the results of one group's method were corroborated by a second group using the same technique. In the present series, we evaluated the results of reconstruction of the extensor mechanism with use of allograft according to the method described by Emerson et al.Methods: Seven reconstructions of the extensor mechanism with use of a bone-tendon-bone allograft were performed with the technique of Emerson et al. in six patients. The patients were evaluated before and after the operation. The knee score according to the system of The Hospital for Special Surgery, evidence of an extensor lag, use of walking aids, and the ambulatory status of each patient were recorded. The patients were also asked about, and the medical records were reviewed for, episodes of falling related to weakness of the quadriceps after the reconstruction. The mean duration of follow-up was thirty-nine months (range, six to 115 months). As these reconstructions often fail early, the minimum duration of follow-up was six months.Results: All seven reconstructions were rated as clinical failures on the basis of a persistent or recurrent extensor lag of more than 30 degrees. All but one patient needed an assistive device full time for walking, and four patients (five knees) had at least one documented episode of falling that was due to giving-way of the affected knee. Four of the reconstructions were revised; one revision was performed with use of another extensor mechanism allograft and three were performed with use of a medial gastrocnemius rotation flap. The other three clinical failures had not been revised at the time of writing. At the time of the most recent follow-up (or at the time of revision of the extensor reconstruction), the mean extensor lag was 59 degrees and the mean knee score was 52 points (a poor result).Conclusions: Undertensioning of the allograft reconstruction at the time of the operation and attenuation of the allograft both may have played a role in the inability of the patients to regain active extension of the knee postoperatively. Alternative techniques for reconstruction of the extensor mechanism or modifications of this technique should be considered in the treatment of this difficult problem.