Eighty-six patients with 89 infected total knee arthroplasties were treated with insertion of another prosthesis. Treatment was not according to an established protocol for parameters, such as delay between removal of the infected prosthesis and insertion of the new prosthesis, duration of antibiotics, use of antibiotic-impregnated cement spacers or beads, and use of antibiotic-impregnated cement for prosthetic fixation at revision surgery. Final followup averaged 52 months (range, 6-126 months). Complications occurred in 30 (33.7%) knees, with recurrent deep infection developing in 10 (11.24%) knees. Patient age, medical diagnosis, type of microorganism, duration of parenteral antibiotics, delay between the resection and revision surgery, and use of antibiotic-impregnated cement spacers or beads were not correlated with the cure rate of infection. Use of antibiotic-impregnated bone cement for prosthesis fixation at revision surgery was the only variable that correlated with the cure rate of deep infection. Seven (28%) of the 25 knees without antibiotic-impregnated cement for prosthesis fixation developed recurrent infection compared with 3 (4.7%) of 64 knees with antibiotic-impregnated cement for prosthesis fixation. This difference was statistically significant (p = 0.0025, log-rank test).