OBJECTIVES: To determine the infection rate and identify the risk factors of primary total knee replacement in a general hospital and discuss possible preventive measure. DESIGN: Retrospective study. SETTING: Regional hospital, Hong Kong. PATIENTS: All cases of primary total knee replacement performed between the period July 1997 and June 2006 were reviewed. MAIN OUTCOME MEASURES: Infection rate of primary total knee replacement and its relationship to risk factors. RESULTS: In the defined period, 479 total knee replacements were performed in 353 patients (291 female and 62 male); 105 women and 21 men had bilateral replacements. The mean patient age was 69 (range, 40-88) years. In all, 447 knees had osteoarthritis, and 32 had rheumatoid arthritis. The mean follow-up period was 46 (range, 1-107) months; 345 knees were followed up longer than 24 months, but seven had no postoperative follow-up. Wound infection was defined by clinical, bacteriological, and/or histological examination. Primary total knee replacement was invariably performed in a theatre with vertical laminar flow, under prophylactic antibiotic cover, and body exhaust suits, water impermeable gowns, and double gloves were always used. The overall infection rate was 3.0% (14/472); the acute deep infection rate (within 4 weeks) was 0.2% (1/472), the delayed deep infection rate (4 weeks-2 years) was 0.6% (2/345). The superficial infection rate was 1.9% (9/472) and the late deep infection rate (after 2 years) was 0.6% (2/345). Diabetic patients had a three-fold higher risk of infection than non-diabetic patients, though this difference did not attain statistical significance (P=0.077). CONCLUSIONS: Our infection rates for primary total knee replacement were comparable to those encountered internationally.