In a prospective study of 14 consecutive infected total knee arthroplasties ( TKAs ) treated through 1979, the management consisted of: (1) 11 delayed exchange arthroplasties, with 2 failures requiring above-knee amputation; (2) 2 in situ debridements; and (3) 1 arthrodesis. The final outcome of the delayed exchange arthroplasty group regarding joint pain, patient function, and joint performance was (a) inferior to that of primary TKA (before infection), (b) better than that of resection arthroplasty, and (c) worse than that of a control group of non-infected TKAs respectively. Debridement without prosthesis resection was successful only in cases of immediate postoperative infection (2 successes in 4 TKAs ) and uniformly unsuccessful in infections occurring beyond the perioperative period (4 failures in 4 TKAs ). The study indicates that delayed exchange total knee arthroplasty is a reasonable alternative to arthrodesis following resection of an infected TKA and can be performed without undue risk of recurrent infection.