A consecutive series of revision total knee arthroplasties performed at 3 university-affiliated centers by 3 surgeons was prospectively studied. The same implant was used in all cases. The evaluation included a Knee Society clinical score (KSCS); SF-36; satisfaction survey; and radiographs preoperatively, at 6 and 12 months postoperatively, and annually thereafter. Follow-up averaging 36 months (range, 24-60 months) was obtained in 125 of 138 knees (91%). Twenty-eight knees were infected, and 26 of 28 knees were treated successfully with 2-stage exchange with an interval of 4 to 6 weeks using an antibiotic-impregnated spacer block and intravenous antibiotics. The remaining 99 knees were revised for reasons other than infection, including aseptic component loosening, progressive osteolysis, and component instability. Preoperatively, patients with infection had a significantly decreased arc of motion compared with patients without infection (79°vs 92°; P <.05). There was a strong trend for the infected knees to have a lower preoperative KSCS than the noninfected knees, although this trend did not achieve statistical significance (76 vs 92; P =.11). Postoperatively, patients with infection continued to have a significantly decreased range of motion (89°vs 99°; P =.05). The postoperative KSCS was markedly lower in the septic versus aseptic revisions (115 vs 135; P =.02). Patients with infection had a significantly lower function score (44 vs 57; P =.03). A significantly higher percentage of patients stated that they were unable to return to normal activities of daily living after septic versus aseptic revision total knee arthroplasty (24% vs 7%; P <.05). Despite the inferior functional result, patients expressed an equal degree of satisfaction with the results of their treatment in septic versus aseptic revision cases.