We analysed the serum C-reactive protein level, synovial fluid obtained by joint aspiration and five synovial biopsies from 145 knee replacements prior to revision to assess the value of these parameters in diagnosing late peri-prosthetic infection. Five further synovial biopsies were used for histological analysis. Samples were also obtained during the revision and incubated and analysed in an identical manner for 14 days. A total of 40 total knee replacements were found to be infected (prevalence 27.6%). The aspiration technique had a sensitivity of 72.5% (95% confidence interval (CI) 58.7 to 86.3), a specificity of 95.2% (95% CI 91.2 to 99.2), a positive predictive value of 85.3% (95% CI 73.4 to 100), a negative predictive value of 90.1% (95% CI 84.5 to 95.7) and an accuracy of 89%. The biopsy technique had a sensitivity of 100%, a specificity of 98.1% (95% CI 95.5 to 100), a positive predictive value of 95.2% (95% CI 88.8 to 100), a negative predictive value of 100% and an accuracy of 98.6%. C-reactive protein with a cut-off-point of 13.5 mg/l had a sensitivity of 72.5% (95% CI 58.7 to 86.3), a specificity of 80.9% (95% CI 73.4 to 88.4), a positive predictive value of 59.2% (95% CI 45.4 to 73.0), a negative predictive value of 88.5% (95% 81.0 to 96.0 CI) and an accuracy of 78.1%. We found that biopsy was superior to joint aspiration and C-reactive protein in the diagnosis of late peri-prosthetic infection of total knee replacements.