Accurate preoperative diagnosis of prosthetic joint infection (PJI) in the absence of obvious clinical signs or laboratory findings is challenging. Interleukin 6 (IL-6) has been proposed as an infection marker but supportive data are limited. We studied the diagnostic utility of serum IL-6 in infected total knee arthroplasty (TKA). A prospective cohort study was done in 52 patients (59 knees) with a painful TKA. The abnormal limits for serum IL-6, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), synovial fluid white cell counts (WBC) and synovial fluid neutrophils (PMN) were determined from receiver operating characteristic curves. An infection was defined as the presence of sinus tract or positive culture > two separate tissue or fluid samples. We utilized Mann-Whitney test, Spearman's correlation and Fischer's exact test to determine the sensitivity, specificity, positive predictive value, negative predictive value (NPV) and accuracy of serum IL-6. The optimal threshold concentration of serum IL-6 was 9.14pg/ml. Independently, this yielded a sensitivity, specificity and NPV of 81%, 63% and 85%, respectively, and when combined with synovial fluid WBC, values were 100%, 90%, 100%, respectively. The sensitivity and specificity of ESR (70%, 63.6%), CRP (66.7%, 66.7%), synovial WBC (66.7%, 81%) and synovial PMN (82.4%, 73.7%) were also calculated. Serum IL-6 levels strongly correlate with all markers in PJI. Serum IL-6 improves the diagnosis of PJI over existing methods, especially when combined with synovial fluid WBC. Its optimal usage is as an excellent screening test to rule out infected total knee arthroplasty. Diagnostic Level II.