The preoperative differentiation between septic and aseptic loosening after total hip or knee arthroplasty is essential for successful therapy and relies in part on biomarkers. The objective of this study was to assess synovial and serum levels of inflammatory proteins as diagnostic tool for periprosthetic joint infection and compare their accuracy with standard tests. 120 patients presenting with a painful knee or hip endoprosthesis for surgical revision were included in this prospective trial. Blood samples and samples of intraoperatively acquired joint fluid aspirate were collected. White blood cell count, C-reactive protein, procalcitonin and interleukin-6 were determined. The joint aspirate was analyzed for total leukocyte count and IL-6. The definite diagnosis of PJI was determined on the basis of purulent synovial fluid, histopathology and microbiology. IL-6 in serum showed significantly higher values in the PJI group as compared to aseptic loosening and control, with specificity at 58.3% and a sensitivity of 79.5% at a cut-off value of 2.6 pg/ml. With a cut-off >6.6 pg/ml, the specificity increased to 88.3%. IL-6 in joint aspirate had, at a cut-off of >2100 pg/ml, a specificity of 85.7% and sensitivity of 59.4%. At levels >9000 pg/ml, specificity was almost at 100% with sensitivity just below 50%, so PJI could be considered proven with IL-6 levels above this threshold. Our data supports the published results on IL-6 as a biomarker in PJI. In our large prospective cohort of revision arthroplasty patients, the use of IL-6 in synovial fluid appears to be a more accurate marker than either the white blood cell count or the C-reactive protein level in serum for the detection of periprosthetic joint infection. On the basis of the results we recommend the use of the synovial fluid biomarker IL-6 for the diagnosis of periprosthetic joint infection following total hip and knee arthroplasty.