Objective To investigate the effectiveness of two-stage revision of infected total knee arthroplasty (TKA) using an antibiotic-impregnated articulating cement spacer. Methods The clinical data were analyzed from 23 patients (23 knees) undergoing two-stage revision for late infection after primary TKA between January 2007 and December 2009. There were 15 males and 8 females, aged from 43 to 75 years (mean, 65.2 years). Infection occurred at 13-52 months (mean, 17.3 months) after TKA. The time interval between infection and admission ranged from 15 days to 7 months (mean, 2.1 months). One-stage operation included surgical debridement and removal of all knee prosthesis and cement, then an antibiotic-impregnated articulating cement spacer was implanted. The re-implantation of prosthesis was performed after 8-10 weeks when infections were controlled. The American Hospital for Special Surgery (HSS) score and Knee Society Score (KSS) were used to compare the function of the knee between pre- and post-revision. The rate of infection control and complication were analyzed. Results All incisions healed primarily. Re-infection occurred in 2 cases after two-stage revision, and infection was controlled in the other 21 cases, with an infection control rate of 91.3%. The patients were followed up 2-5 years (mean, 3.6 years). The HSS score was increased from 60.6 ± 9.8 at pre-revision to 82.3 ± 7.4 at last follow-up, the KSS score was increased from 110.7 ± 9.6 at pre-revision to 134.0 ± 10.5 at last follow-up, all showing significant differences (P lt; 0.01). Radiographs showed that prosthesis had good position with no loosening, fracture, or periprosthetic radiolucent. Conclusion Two-stage revision using an antibiotic-impregnated articulating cement spacer is an effective method to control infected TKA and to restore the function of affected knee.